Dynamic risk monitoring system and method based on safety production

Abstract

The invention discloses a risk dynamic monitoring system and method based on safety production, relates to the technical field of safety production, and solves the technical problem of low safety production risk dynamic monitoring precision caused by neglecting of mutual influence of data corresponding to production links in the prior art. According to the method, the whole production process is divided into a plurality of production links, independent analysis of safety production risks is carried out on each production link, and once a certain production link is abnormal, an abnormal reason and an abnormal point can be quickly positioned, so that workers can conveniently process in time; the safety monitoring data of each production link is reasonably predicted based on the historical record data, whether the safety production risk exists or not is judged according to the predicted safety monitoring data, the mutual influence between the production links is fully considered in the reasonable prediction process, the safety monitoring data prediction reasonability can be improved, and the safety production risk is reduced. And thus, the accuracy of safety production risk judgment in each production link is improved.

Description

Technical Field

[0001] The present application belongs to the technical field of safe production, and specifically is a risk dynamic monitoring system and method based on safe production. Background Art

[0002] With the acceleration of industrialization and rapid economic development, production safety issues have gradually revealed their complexity and diversity. Especially in high-risk production industries, once a safety accident occurs, it will not only cause economic losses to the enterprise, but also have a significant impact on personnel safety and the environment.

[0003] Many processes in current risk assessment methods are based on expert opinions or manual assessments, which greatly affect the assessment results by personal experience. In addition, the assessment process takes a certain amount of time, making it difficult to ensure the accuracy and timeliness of the assessment results. There are also some methods that mine and analyze data related to production safety through various models to give assessment results. This type of method can improve data processing efficiency, but due to the large amount of data processed and the variety of data types, it is difficult to sort out the multidimensional relationship between the various types of data, ignoring the mutual influence between the data, and still cannot guarantee the accuracy of the assessment results.

[0004] The present application provides a risk dynamic monitoring system and method based on safe production to solve the above technical problems. Summary of the invention

[0005] The present application aims to solve at least one of the technical problems existing in the prior art; to this end, the present application proposes a risk dynamic monitoring system and method based on production safety, which is used to solve the technical problem that the prior art predicts the occurrence of production safety risks through a single type of data change trend, ignoring the mutual influence of data corresponding to production links, resulting in low accuracy of dynamic monitoring of production safety risks.

[0006] To achieve the above-mentioned purpose, the first aspect of the present application provides a risk dynamic monitoring system based on safe production, including a data processing layer, and a data collection layer for data interaction therewith;

[0007] Data collection layer: used to divide the production process into several production links, and collect safety monitoring data of several production links in real time through the set data sensors; wherein the data sensors are set in accordance with the safety production regulations, and several production links are related;

[0008] Data processing layer: used to extract historical record data of each production link from the database, identify the mutual influence relationship between the production links according to the historical record data, and determine the associated influence data; wherein the associated influence data includes associated influence factors and corresponding associated influence weights; and,

[0009] The safety monitoring data of the production link is adjusted based on the associated impact data and marked as target monitoring data; the safety risk of the corresponding production link is determined based on the target monitoring data.

[0010] Preferably, identifying the mutual influence relationship between production links based on historical record data includes:

[0011] Select the production link as the target link and other production links as the associated links in turn; set the time interval based on the association relationship between the target link and the associated links; set the time 1 and the time 2 according to the time interval and the initial data collection time;

[0012] Extracting a number of data groups corresponding to the target link from the historical record data based on time one and time two; wherein the data groups include the historical record data of the target link at time one and time two, and the historical record data of the associated link at time one;

[0013] Identify the mutual influence between production links based on several data sets.

[0014] Preferably, extracting several data groups corresponding to the target link from the historical record data based on time one and time two includes:

[0015] Based on time one, data sequence one is obtained by matching the historical record data of the target link, and based on time two, data sequence two is obtained by matching the historical record data of the associated link;

[0016] Integrate data sequence 1 and data sequence 2 into a data group; update time 1 and time 2 sequentially based on a time interval, and rematch and integrate the data to obtain a number of data groups.

[0017] Preferably, the mutual influence relationship between production links is identified based on a plurality of data groups, including:

[0018] Extract the data difference between the historical record data corresponding to the target link time 1 and time 2 in the data group as the training output data, and associate the historical record data corresponding to the link time 1 as the training input data;

[0019] The neural network model is trained by using preprocessed training input data and training output data, and the associated influence data corresponding to the target link is extracted from the trained neural network model; wherein the neural network model includes at least one hidden layer.

[0020] Preferably, extracting the associated impact data corresponding to the target link from the trained neural network model includes:

[0021] Calculate the total weight of each type of data in the historical record data according to the trained neural network model corresponding to the target link, and select and determine several related influencing factors according to the total weight;

[0022] The total weights of several associated influencing factors are normalized to obtain associated influencing weights of several associated influencing factors; and the several associated influencing factors and the corresponding associated influencing weights are integrated into associated influencing data of the target link.

[0023] Preferably, the total weight of each type of data in the historical record data is calculated according to the trained neural network model corresponding to the target link, including:

[0024] Determine several data types in the historical record data, and extract weight matrices corresponding to the several data types from the neural network model;

[0025] Take the absolute values ​​of the weights in the weight matrix and sum them up to get the total weight of the data type.

[0026] Preferably, adjusting the safety monitoring data of the production process based on the associated impact data includes:

[0027] Determine the change trend of the safety monitoring data corresponding to the target link through the associated impact data corresponding to the target link and the safety monitoring data of the associated link; wherein the change trend includes the change direction and the change degree;

[0028] The target monitoring data is obtained by superimposing the safety monitoring data of the target link and its change trend.

[0029] Preferably, after obtaining the target monitoring data, the weight coefficient of each data type in the target link can also be set;

[0030] The safety risk assessment value is calculated based on the weight coefficient and target monitoring data, and whether there is a safety risk in the target link is determined based on the safety risk assessment value.

[0031] Preferably, judging the safety risks of the corresponding production links based on the target monitoring data includes:

[0032] Set the monitoring data threshold corresponding to the production stage;

[0033] When any type of data in the target monitoring data is greater than the corresponding monitoring data threshold, it is determined that there is a security risk and an early warning is issued based on the determination result.

[0034] The second aspect of the present application provides a risk dynamic monitoring method based on safe production, comprising:

[0035] The production process is divided into several production links, and the safety monitoring data of several production links are collected in real time through the data sensors set up; wherein the data sensors are set up in accordance with the safety production regulations, and the several production links are related;

[0036] Extracting historical record data of each production link from the database, identifying the mutual influence relationship between the production links according to the historical record data, and determining the associated influence data; wherein the associated influence data includes associated influence factors and corresponding associated influence weights;

[0037] The safety monitoring data of the production link is adjusted based on the associated impact data and marked as target monitoring data; the safety risk of the corresponding production link is determined based on the target monitoring data.

[0038] Compared with the prior art, the beneficial effects of this application are:

[0039] 1. In this application, the entire production process is first divided into multiple production links, and the safety production risks of each production link are analyzed separately. Once a production link is abnormal, the cause and abnormal point of the abnormality can be quickly located, which is convenient for the staff to deal with it in time; the safety monitoring data of each production link is reasonably predicted based on the historical record data, and whether there is a safety production risk is judged according to the predicted safety monitoring data. In the reasonable prediction process, the mutual influence between the production links is fully considered, which can improve the rationality of the safety monitoring data prediction, and then improve the accuracy of the safety production risk judgment of each production link; the corresponding related influence data is mined for each production link, which is more targeted and conducive to improving the safety risk monitoring accuracy of a single production link.

[0040] 2. The present application provides a method for acquiring several data groups corresponding to a target link, wherein time one and time two are determined according to a set time interval and an initial collection time, data groups are extracted from data groups according to preset rules through time one and time two, and time one and time two are updated according to the time interval to obtain several data groups corresponding to the target link; if data of time one or time two does not exist in the historical record data, data of the previous and next collection times are used to supplement the data to improve the reliability and accuracy of the several data groups, thereby ensuring the accuracy of the associated influencing data corresponding to the target link. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0042] Figure 1 This is a schematic diagram of the method flow of the risk dynamic monitoring system in Example 1 of the present application;

[0043] Figure 2 This is a schematic diagram of the system principle of the risk dynamic monitoring system in Example 1 of the present application;

[0044] Figure 3 This is a schematic diagram of the determination process of time 1 and time 2 in Example 1 of the present application;

[0045] Figure 4 This is a schematic diagram of the process of determining several data groups corresponding to the target link in Example 1 of the present application. DETAILED DESCRIPTION

[0046] The technical solution of the present application will be described clearly and completely in conjunction with the embodiments below. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application.

[0047] The importance of safe production is self-evident, but there are still many problems in the early warning and monitoring methods of safe production. During the entire production process of the product, it needs to go through many production links for processing, and it also requires the cooperation of many staff. In each production link, either set the relevant data threshold through expert experience, compare the production data collected by the corresponding production link with the data threshold, and determine whether the production data is abnormal for early warning; or use an artificial intelligence model to mine the data relationship in the historical record data of the production link, and use the data relationship to determine whether the production data is abnormal for early warning. Most of the existing solutions focus on the separate analysis of each production link, but in fact the data of other production links will have an impact on the data of the analyzed production link. If this part of the impact is not considered, it will obviously affect the accuracy of the final analysis results. In order to solve the above problems, this application provides the following technical solutions.

[0048] Embodiment 1:

[0049] See also Figure 1-Figure 4 , the first aspect of the present application provides a risk dynamic monitoring system based on safe production, including a data processing layer, and a data collection layer for data interaction therewith;

[0050] Data collection layer: used to divide the production process into several production links, and collect safety monitoring data of several production links in real time through the set data sensors; wherein the data sensors are set in accordance with the safety production regulations, and several production links are related;

[0051] Data processing layer: used to extract historical record data of each production link from the database, identify the mutual influence relationship between the production links based on the historical record data, and determine the associated influence data; wherein the associated influence data includes associated influence factors and corresponding associated influence weights; and, based on the associated influence data, adjust the safety monitoring data of the production link and mark it as target monitoring data; judge the safety risk of the corresponding production link based on the target monitoring data.

[0052] In this embodiment, the data collection layer is used to divide the production process into multiple production links, and then collect the safety monitoring data of each production link through the set data sensors, and exchange data with the data processing layer.

[0053] The production process of a product contains many steps, which can be divided according to the order of the production process, or according to the spatial position of the equipment in the production process. The division of the production process is to achieve refined monitoring and early warning, so as to facilitate the location of abnormalities. The data sensors set up are mainly used to collect safety monitoring data of each production link. The safety monitoring data mainly includes the status data of production equipment, staff, sources of danger, etc. These data are closely related to the safety of the production link. It should be noted that data sensors include both additional sensors and sensors set inside the equipment, and data sensors do not need to be set up separately for each production link. It is only necessary to ensure that the data collected by the data sensor can be used in the corresponding production link.

[0054] In this embodiment, the data processing layer is mainly used for data analysis and monitoring and early warning, that is, judging whether there are safety risks in each production link based on historical record data and safety monitoring data.

[0055] The database is connected to the data processing layer so that the data processing layer can extract the required data from the database at any time. The database stores the historical record data corresponding to each production link. Based on these historical record data, the mutual influence and degree of mutual influence between each production link can be judged, thereby providing a basis for risk warning of each production link. The historical record data includes the recorded safety monitoring data and the corresponding safety production status, and the real-time collected safety monitoring data will also be saved in the database in time, and the historical record data will be updated for subsequent use.

[0056] After determining the mutual influence between each production link, the mutual influence and the real-time collected safety monitoring data can be combined to analyze whether there is a safety production risk in any production link, so as to provide early warning for safety production risks. The introduction of the mutual influence between production links can improve the accuracy of safety production risk judgment and the accuracy of overall risk monitoring.

[0057] It is worth noting that the mutual influence between production links refers to the influence of other production links on all or part of the data of the target link, such as the influence of the speed adjustment of the previous production link on the speed of the next production link. Moreover, there will be a certain delay in the actual effect of the mutual influence of each production link, and this delay is an important parameter for safety production risk prediction. Moreover, it is not only the previous production link that has an impact on the next production link, but the next production link will also have an impact on the previous production link. Many aspects need to be considered when introducing mutual influence. It should be noted that when exploring the mutual influence relationship, it mainly includes the influence between the equipment status parameters corresponding to each production link in the production process, and the focus is on judging whether the equipment of each production link may have risks in the production process. Of course, if there are other data that influence each other in the production process of the product, the technical solution provided by this application can also be used.

[0058] Explanatory, the data types mentioned in this application can be divided according to content, such as equipment data, personnel data, hazard source data, etc.; they can also be divided according to data type, such as temperature data, pressure data, oil level data, etc. Of course, other data division methods that are helpful for safety risk analysis can also be referenced in this solution.

[0059] Next, we will explain in detail how to determine the mutual influence between various production links, that is, to identify the mutual influence relationship between production links based on historical record data. Figure 3 ,include:

[0060] Select production links as target links and other production links as associated links in turn; set time intervals based on the association between the target links and the associated links; set time one and time two according to the time interval and the initial data collection time; extract several data groups corresponding to the target links from the historical record data based on time one and time two; identify the mutual influence relationship between the production links based on several data groups.

[0061] In the actual production process, each production link may be affected by other production links, so the production links are selected as target links in turn. When a production link is the target link, other production links are related links. As mentioned above, there is a delay in the mutual influence between the production links, that is, the time interval is set according to the delay, which is very important for subsequent data extraction.

[0062] Time 1 and time 2 are set according to the time interval and the initial data collection time. The initial data collection time can be understood as the collection time of the first data in the historical record data. This collection time is taken as time 1, and the time after the aforementioned time interval is taken as time 2. In order to completely extract the historical record data, the aforementioned time 2 can be taken as time 1, and a new time 2 can be determined according to the time interval.

[0063] It is worth noting that although the production links may affect each other, the delay corresponding to the mutual influence is not fixed, that is, the time interval between the target link and each related link may be different, which will also affect the subsequent extraction of data groups from historical record data.

[0064] According to the determined time 1 and time 2, several data groups corresponding to each target link are extracted from the historical record data, and the mutual influence relationship with other related links is determined through the several data groups corresponding to the target link. In this way, the target links are determined in turn, and after analyzing each target link, the mutual influence relationship between the production links in the entire production process can be obtained.

[0065] The existence of a temporal relationship between several production links means that some production links have a continuous relationship between the previous and the next production link, such as the next production link can only be carried out after the previous production link is completed; the existence of a temporal relationship between several production links means that there is no temporal relationship between the two, such as the next production link C can be carried out after production link A and production link B are completed, then the relationship between A and B is not a temporal relationship.

[0066] Next, we will give several data sets corresponding to each target link according to time 1 and time 2. Figure 4 , please refer to the following steps for details:

[0067] Based on time one, data sequence one is matched from the historical record data of the target link, and based on time two, data sequence two is matched from the historical record data of the associated link; data sequence one and data sequence two are integrated into a data group; based on the time interval, time one and time two are updated sequentially, and the data are re-matched and integrated to obtain several data groups.

[0068] After determining the target link and time 1 and time 2, extract data sequence 1 from the historical record data of the target link based on time 1, and extract data sequence 2 from the historical record data corresponding to the associated link of the target link based on time 2. Combine data sequence 1 and data sequence 2 into one data group, and due to the update of time 1 and time 2, several data groups corresponding to the target link can be obtained.

[0069] Each data group includes the historical record data of the target link at time 1 and time 2, and the historical record data of the associated link at time 1 (time, time 2 is later than time 1). That is, each data group includes the historical record data of each production link at time 1, and the historical record data of the target link at time 2. Based on these data, it is possible to mine which data in the associated link will have an impact on the target link and the extent of the impact.

[0070] It is worth noting that this embodiment divides the historical record data of each production link in time by setting time 1 and time 2 to obtain the data group corresponding to the target link. However, in fact, many types of data in the historical record data are independent and will not be affected by changes in data related to other production links, nor will they affect the relevant data of other production links, such as the number of staff. For these independent data, they can be marked through preliminary correlation analysis, and these data can be eliminated when generating data groups to reduce the amount of data processing.

[0071] Moreover, by extracting several data groups corresponding to the target link at time one and time two, each data group contains the influence of the recent data changes of the associated link on the various types of data in the target link, and each data group implies an influence relationship. By synthesizing the influence relationships of several data groups, we can obtain the degree of change of the target link data caused by the degree of change of different data in the associated link. The existing scheme directly predicts based on the data of the target link, fixes the data change trend of the target link, and actually fixes the influence of the associated link on the data change trend of the target link. However, if the data of the associated link at time one changes abnormally, its influence on the data change trend of the target link will also change unpredictably. Therefore, the method provided by the present application to determine several data groups based on time one and time two to determine the mutual influence relationship of each production link will make a more refined dynamic prediction of the safety risk of the target link based on the data collected in real time, with higher accuracy and better effect.

[0072] After obtaining several data groups corresponding to each target link, the associated impact data can be extracted through the neural network model. The principle of this method is that when the neural network model is trained with accurate historical record data, the impact of the associated link data on the change of the target link data will be continuously learned during the training process, that is, the weight distribution will be continuously optimized to build the mapping relationship between the model input data and the model output data. When the mapping relationship is built, the weight distribution has reached the optimal condition, and the associated impact data can be extracted according to the assigned weight.

[0073] The data difference corresponding to the target link at time 1 and time 2 is extracted from the data groups corresponding to the target link as the training output data. The data difference refers to the change between various types of data. The data difference may be one data or multiple data. The specific number is determined according to the data type in the historical record data. At the same time, the historical record data corresponding to time 1 of the associated link is extracted from the data groups as the training input data.

[0074] Before training the neural network model, the training input data and training output data need to be preprocessed. This preprocessing is similar to the data preprocessing in the existing model training process, and the detailed preprocessing process is not repeated here. It is also necessary to determine the input layer and output layer according to the number of data types in the training input data and training output data, and include at least one hidden layer as a weight optimization layer.

[0075] After the neural network model is constructed, the neural network model is trained through the preprocessed training input data and training output data, and the trained neural network model is obtained after evaluation. The optimized weights can be extracted from the hidden layer of the neural network model. Please refer to the following steps:

[0076] The total weight of each type of data in the historical record data is calculated according to the trained neural network model corresponding to the target link, and a number of associated influencing factors are screened and determined according to the total weight; the total weight of the several associated influencing factors is normalized to obtain the associated influence weights of the several associated influencing factors; the several associated influencing factors and the corresponding associated influence weights are integrated into the associated influence data of the target link.

[0077] The total weight is determined as follows:

[0078] Determine several data types in the historical record data, extract weight matrices corresponding to the several data types from the neural network model; take the absolute values ​​of the weights in the weight matrix and sum them up to obtain the total weight of the data type.

[0079] Exemplarily, the types of input data of the neural network model are A1, A2, and A3, and the types of output data are B1 and B2. For B1 and B2, the total weights corresponding to A1, A2, and A3 can be extracted from the trained neural network model. If for B1, the sum of the total weights of A1 and A2 is very large, such as 99%, the influence of A3 on B1 can be ignored. After removing A1, the total weights of A1 and A2 are normalized to obtain the associated influence data. If for B2, the total weights of A1, A2, and A3 are not much different, then the total weights of A1, A2, and A3 can be directly normalized.

[0080] After determining the associated impact data of the target link, the security monitoring data corresponding to the target link can be extracted, and each data type in the security monitoring data can be associated with the corresponding associated impact data. The predicted change amount of each type of data in the target link is determined by combining the collected security monitoring data and associated impact data of other associated links, and the predicted change amount is superimposed with the security monitoring data corresponding to the current target link to obtain the predicted security monitoring data, that is, the target monitoring data.

[0081] It should be noted that the target monitoring data can be calculated using a trained neural network model, which can improve the calculation accuracy, but the neural network model needs to be updated and stored in a timely manner, which will increase the cost; the associated impact data extracted from the neural network model can also be directly associated with the corresponding production link, and can be directly extracted and calculated when calculation is required, which can effectively reduce the cost; but in order to ensure the reliability of the calculation results, the associated impact data needs to be updated according to the newly collected historical record data. This embodiment implements the technical solution of the present application in the second way.

[0082] After obtaining the target monitoring data of the production link, if a certain type of data in the target monitoring data is greater than the preset monitoring data threshold, it is determined that there is a safety risk in the production link, and an early warning signal is generated to remind the staff to deal with it.

[0083] Of course, it is also possible to preset weights for each type of data in the target monitoring data, calculate a safety risk assessment value based on the preset weights and each type of data, compare the safety risk assessment value with the set risk assessment threshold, and determine whether there is a safety production risk in the corresponding production link. It should be noted that if each type of data in the target monitoring data is not greater than the preset monitoring data threshold, the safety risk assessment value is also calculated using the preset weights to determine whether the combined effect of each type of data may affect the safety production of the production link.

[0084] In this embodiment, the entire production process is first divided into multiple production links, and the safety production risk is analyzed separately for each production link. Once a production link is abnormal, the cause and abnormal point of the abnormality can be quickly located, which is convenient for the staff to deal with it in time; based on the historical record data, the safety monitoring data of each production link is reasonably predicted, and it is judged whether there is a safety production risk based on the predicted safety monitoring data. In the reasonable prediction process, the mutual influence between production links is fully considered, which can improve the rationality of the safety monitoring data prediction, and then improve the accuracy of the safety production risk judgment of each production link.

[0085] Embodiment 2: Based on determining the time 1 and the time 2 according to the method provided in Embodiment 1, a method for extracting several data groups from the historical record data according to the time 1 and the time 2 is provided, as shown below:

[0086] Time 1 is determined according to the initial collection time corresponding to the historical record data, and time 2 is equal to time 1 plus the set time interval. However, due to the limitations of the time interval setting, when extracting historical record data based on time 1 or time 2, the data at the corresponding time may be empty.

[0087] In order to ensure that enough data sets can be extracted and the reliability of the data sets can be guaranteed as much as possible, if the data at the corresponding moment is empty, the data closest to it will be extracted to make up for it. If there is data before and after the corresponding moment and the time difference is almost the same, the average of the data before and after will be taken to make up for it.

[0088] Embodiment 3: Based on Embodiment 1, a method for generating training input data and training output data when training a neural network model is provided, as shown below:

[0089] As shown in Example 1, the function of the neural network model of the present application is to establish the influence relationship of each type of data in the associated link on each type of data in the target link when the data changes. Therefore, the training input data at least includes the current change amount of each type of data, which is the difference between the current value of each type of data and the value at the previous acquisition time. Similarly, the training output data also corresponds to the change amount of each type of data. The above change amount can be calculated from the data in the historical record data.

[0090] It should be noted that the current change in the training input data can be understood as the value at moment one minus the value at the previous acquisition moment (if there is no value at the previous acquisition moment, this group of data can be eliminated or supplemented with the data at moment one), and the change in the training output data can be understood as the value at moment two minus the value at moment one.

[0091] The second aspect of the present application provides a method for dynamic risk monitoring based on safe production, including:

[0092] The production process is divided into several production links, and the safety monitoring data of several production links are collected in real time through the data sensors set up; wherein the data sensors are set up in accordance with the safety production regulations, and the several production links are related;

[0093] Extracting historical record data of each production link from the database, identifying the mutual influence relationship between the production links according to the historical record data, and determining the associated influence data; wherein the associated influence data includes associated influence factors and corresponding associated influence weights;

[0094] The safety monitoring data of the production link is adjusted based on the associated impact data and marked as target monitoring data; the safety risk of the corresponding production link is determined based on the target monitoring data.

[0095] The above embodiments are only used to illustrate the technical method of the present application and are not intended to limit it. Although the present application has been described in detail with reference to the preferred embodiments, a person of ordinary skill in the art should understand that the technical method of the present application may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present application.

Claims

1. A risk dynamic monitoring system based on production safety, characterized in that: It includes the data processing layer and the data collection layer that interacts with it; Data collection layer: used to divide the production process into several production links, and collect safety monitoring data of several production links in real time through the set data sensors; wherein the data sensors are set in accordance with the safety production regulations, and several production links are associated with each other; Data processing layer: used to extract historical record data of each production link from the database, identify the mutual influence relationship between the production links according to the historical record data, and determine the associated influence data; wherein the associated influence data includes associated influence factors and corresponding associated influence weights; and, The safety monitoring data of the production link is adjusted based on the associated impact data and marked as target monitoring data; and the safety risk corresponding to the production link is determined based on the target monitoring data.

2. A risk dynamic monitoring system based on production safety according to claim 1, characterized in that: Identifying the mutual influence relationship between the production links according to the historical record data includes: Selecting the production link as the target link and other production links as associated links in sequence; setting a time interval based on the association relationship between the target link and the associated links; setting time 1 and time 2 according to the time interval and the initial data collection time; Extracting a plurality of data groups corresponding to the target link from the historical record data based on the moment 1 and the moment 2; wherein the data groups include the historical record data of the target link at the moment 1 and the moment 2, and the historical record data of the associated link at the moment 1; Based on the plurality of data groups, the mutual influence relationships between the production links are identified.

3. A risk dynamic monitoring system based on production safety according to claim 2, characterized in that: Extracting a plurality of data groups corresponding to the target link from the historical record data based on the moment 1 and the moment 2, including: Based on the moment one, a data sequence one is obtained by matching the historical record data of the target link, and based on the moment two, a data sequence two is obtained by matching the historical record data of the associated link; The data sequence 1 and the data sequence 2 are integrated into a data group; the moment 1 and the moment 2 are updated sequentially based on the time interval, and the data are re-matched and integrated to obtain a plurality of data groups.

4. A risk dynamic monitoring system based on production safety according to claim 2, characterized in that: Identifying the mutual influence relationship between the production links based on the plurality of data groups includes: Extracting the data difference between the historical record data corresponding to the target link time 1 and time 2 in the data group as training output data, and associating the historical record data corresponding to the link time 1 as training input data; The neural network model is trained by using the preprocessed training input data and the training output data, and the associated influence data corresponding to the target link is extracted from the trained neural network model; wherein the neural network model includes at least one hidden layer.

5. A risk dynamic monitoring system based on production safety according to claim 4, characterized in that: The associated impact data corresponding to the target link is extracted from the trained neural network model, including: Calculate the total weight of each type of data in the historical record data according to the trained neural network model corresponding to the target link, and screen and determine a number of related influencing factors according to the total weight; The total weights of the plurality of associated influencing factors are normalized to obtain associated influencing weights of the plurality of associated influencing factors; and the plurality of associated influencing factors and the corresponding associated influencing weights are integrated into associated influencing data of the target link.

6. A risk dynamic monitoring system based on production safety according to claim 5, characterized in that: The calculating the total weight of each type of data in the historical record data according to the trained neural network model corresponding to the target link includes: Determine several data types in the historical record data, and extract weight matrices corresponding to several of the data types from the neural network model; The absolute values ​​of the weights in the weight matrix are taken and summed to obtain the total weight of the data type.

7. A risk dynamic monitoring system based on production safety according to claim 6, characterized in that: Adjusting the safety monitoring data of the production link based on the associated impact data includes: Determine the change trend of the safety monitoring data corresponding to the target link through the associated impact data corresponding to the target link and the safety monitoring data of the associated link; wherein the change trend includes the change direction and the change degree; The safety monitoring data of the target link and its change trend are superimposed to obtain the target monitoring data.

8. A risk dynamic monitoring system based on production safety according to claim 7, characterized in that: After obtaining the target monitoring data, the weight coefficient of each data type in the target link can also be set; A safety risk assessment value is calculated based on the weight coefficient and the target monitoring data, and whether there is a safety risk in the target link is determined according to the safety risk assessment value.

9. A risk dynamic monitoring system based on production safety according to claim 1, characterized in that: Determining the safety risk corresponding to the production link based on the target monitoring data includes: Setting the monitoring data threshold corresponding to the production link; When any type of data in the target monitoring data is greater than the corresponding monitoring data threshold, it is determined that there is a security risk and an early warning is issued based on the determination result.

10. A method for dynamic monitoring of risks based on safe production, based on the operation of a dynamic monitoring system for risks based on safe production according to any one of claims 1 to 9, characterized in that: include: The production process is divided into several production links, and the safety monitoring data of several production links are collected in real time through the data sensors provided; wherein the data sensors are provided in accordance with the safety production regulations, and the several production links are related; Extracting historical record data of each production link from the database, identifying the mutual influence relationship between the production links according to the historical record data, and determining the associated influence data; wherein the associated influence data includes associated influence factors and corresponding associated influence weights; The safety monitoring data of the production link is adjusted based on the associated impact data and marked as target monitoring data; and the safety risk corresponding to the production link is determined based on the target monitoring data.

Images