Mine transportation safety adaptive optimization system and method based on video monitoring

Abstract

The invention discloses a mine transportation safety adaptive optimization system and method based on video monitoring, and belongs to the technical field of real-time optimization. According to the invention, various sensors are used for sensing the environment in a monitoring area to obtain environment multi-source data in the monitoring area; analyzing and processing the environment sensed by the sensor, and performing feature extraction on the environment in the monitoring area to obtain a first feature of the monitoring area; the environment in the monitoring area is analyzed, the abnormal environment index in the monitoring area is calculated, and graded early warning of abnormity in the monitoring area is achieved; performing machine learning by using the calculated first feature in the monitoring area, and constructing an anomaly early warning model; analyzing the environment during early warning, and extracting a second feature of the monitoring area; linear features of the first features and the second features are obtained through calculation, and real-time optimization of early warning grading is achieved according to the linear features, obtained through calculation, of the first features and the second features.

Description

Technical Field

[0001] The present invention relates to the technical field of real-time optimization, and particularly to an adaptive optimization system and method for mine transportation safety based on video surveillance. Background Technique

[0002] With the advancement of intelligent mines, mine transportation safety has become a core concern. Traditional video surveillance systems are insufficient in dealing with massive data, complex environments, and real-time requirements, and it is difficult to meet the efficient and safe operation needs of modern mines. With the development of science and technology, intelligent video surveillance technology has gradually been applied to mine transportation safety. Intelligent video surveillance technology can monitor various data in the mine transportation scenario by using visual sensors, microphones, temperature and humidity sensors, etc. With the development of modern information technology, intelligent video surveillance plays a crucial role in mine transportation; through the collection and monitoring of various data during mine transportation, a comprehensive investigation of mine transportation safety is carried out to ensure that when abnormal situations occur during mine transportation, it can be promptly reported to the staff for handling; however, due to the complex and changeable working environment of mines, the changes of many data are relatively frequent, and there are also mutual influences between data, resulting in the continuous change of the criteria for abnormal occurrences during mine transportation, leading to insufficient timeliness of early warnings by intelligent video surveillance technology when the mine transportation environment changes and being unable to distinguish the influence relationships between data during mine transportation. Therefore, the present invention designs an adaptive optimization system and method for mine transportation safety based on video surveillance according to the above problems. Summary of the Invention

[0003] The purpose of the present invention is to provide an adaptive optimization system and method for mine transportation safety based on video surveillance to solve the problems raised in the above background technique.

[0004] To solve the above technical problems, the present invention provides the following technical solutions: An adaptive optimization method for mine transportation safety based on video surveillance, the method comprising the following steps: S100. Use a variety of sensors to sense the environment in the monitoring area to obtain multi-source environmental data in the monitoring area; Further, the sensors adopt visual sensors, temperature and humidity sensors, and microphone sensors, and the multi-source environmental data includes temperature and humidity data, image data, and sound data in the monitoring area.

[0005] S200. Process, analyze, and extract features from the environmental data sensed by the sensors to obtain the first feature of the monitoring area; Further, the specific steps for extracting the first feature of the monitoring area by extracting features from the environment in the monitoring area are: S201. Collect the environmental data without anomalies in the monitoring area using multiple sensors, classify them, and divide the data collected by different sensors into different categories according to the sensors. Generate corresponding n types of data, which are respectively , the 1st, 2nd, 3rd... nth types of data respectively, where n is a positive integer; an anomaly in the monitoring area means an explosive change in multiple types of environmental data in the monitoring area.

[0006] S202. Record the environmental data with anomalies in the monitoring area, classify the data, and divide the data collected by different sensors into different categories according to the sensors. Generate m types of abnormal environmental data, which are respectively , the 1st, 2nd, 3rd... mth types of abnormal environmental data respectively, where m is a positive integer; S203. After classifying and collecting the environmental data when there are no anomalies and when there are anomalies in the monitoring area, compare the data in the two cases, and use the environmental data without anomalies in the monitoring area to screen the environmental data with anomalies in the monitoring area , delete the same values in the environmental data with anomalies in the monitoring area as those in the environmental data without anomalies , and define the environmental data with anomalies after deletion as differential data , which are the 1st, 2nd, 3rd... jth types of differential data, where j is a positive integer; S204. According to the above steps, extract the environmental data with anomalies in the monitoring area k times, and then generate k times of differential data after comparing with the environmental data without anomalies in the monitoring area k times. Statistically analyze the k times of differential data, and extract the data type with the most repeated occurrences among the k times of differential data , where the subscript y ranges from 1 to j, and set as the first feature in the monitoring area.

[0007] The first feature generated above reflects the most significant change characteristics in the environmental data when there are anomalies in the monitoring area, and it is possible to judge whether there are anomalies in the monitoring area by monitoring the first feature.

[0008] S300. Analyze the environmental data in the monitoring area, calculate the change speed of the abnormal environmental data in the monitoring area, and realize the hierarchical early warning of the anomalies in the monitoring area; Furthermore, the steps to realize the hierarchical early warning in the monitoring area are as follows: S301. Based on the collection of abnormal environmental data in the monitored area, after k times of comparison between the normal environmental data and the abnormal environmental data in the monitored area in S200, the number of differential data generated k times is , , where and are the numbers of differential data generated by comparing the normal environmental data and the abnormal environmental data in the monitored area for the 1st, 2nd, 3rd... kth times, and k is a positive integer; S302. Sort the first feature values in ascending order, and use the point plotting method to draw a line chart of the first feature values and the number of differential data , where p ranges from 1 to k; S303. Normalize and fit the line chart to form a smooth function curve, and use the segmentation method to calculate the function relationship of each segment in the function curve to generate a function set as , , where and are the functions of the 1st, 2nd, 3rd... hth segments in the function curve, and h is a positive integer; S304. Perform derivative operations on the piecewise functions respectively to obtain the derivative functions of each segment of the function curve, , where and are the derivative functions of the 1st, 2nd, 3rd... hth segments of the function curve, and h is a positive integer; the derivative function represents the growth rate of each segment of the function curve; S305. Screen the derivative functions obtained above, select the derivative functions with values greater than 0 and sort them in ascending order, and obtain the sorting of the corresponding independent variables as , , where and are the independent variables corresponding to the 1st, 2nd, 3rd... a derivative functions greater than 0 sorted by the derivative function values, and a is a positive integer; the calculated is defined as the grading data for early warning; The larger the value of the derivative function of the curve function, the faster the variable changes when the independent variable increases; it can reflect that when an abnormality occurs in the monitored area, by judging the change of the first feature value and the change of the remaining environmental data in the monitored area, the faster the derivative is, the faster the impact of the abnormality in the monitored area is. Therefore, setting the early warning level according to the size of the derivative can effectively classify the speed of the impact of the abnormality in the monitored area on the environment and allocate resources reasonably.

[0009] S306. Perform hierarchical early warning on the monitored area according to the calculated grading data , and select the independent variable The initial value is the minimum level of the warning, _min, and then the warning level is increased sequentially according to the order.

[0010] The above-mentioned hierarchical data of the warning in the monitoring area calculated according to the derivative. After an anomaly occurs in the monitoring area, the speed of increase in the number of environmental data that change due to the change of the first feature will also be different, and the value of the derivative represents the speed of increase in the environmental data that change in the monitoring area. The anomalies are classified and warned according to the different speeds of increase. For example, when monitoring whether a fire may occur during mine transportation, in the monitoring area, let the temperature be the first feature. When the temperature is too high, it is judged that a fire anomaly occurs, and after the fire occurs, the remaining data in the monitoring area will change, and the speed of increase in the number of changed data proves the size of the fire from the side. Therefore, hierarchical warning of anomalies can be achieved.

[0011] S400. Use the first feature calculated in the monitoring area for machine learning to build an anomaly warning model; Furthermore, building the anomaly warning model specifically is: Use the calculated hierarchical data for machine learning using the convolutional neural network algorithm to build a hierarchical warning model; use sensors to perceive the environment in the monitoring area in real time and extract the first feature values in the environmental data in the monitoring area _s, and input the extracted first feature values _s into the built hierarchical warning model. When it is judged that the extracted first feature values _s > _min, it means that the environment in the monitoring area is abnormal. Then compare _s with the hierarchical data in sequence. When it is judged that _s > , issue a warning of the corresponding level to the monitoring area, where i belongs to 1 - a.

[0012] S500. Analyze the environment during the warning and extract the second feature of the monitoring area; Furthermore, the specific steps for extracting the second feature of the monitoring area are: S501. Collect the environmental data with anomalies in the monitoring area, extract the environmental data with anomalies in the monitoring area k times according to the method described in S200, classify each extracted environmental data, and generate m types of abnormal environmental data respectively as , extract the first feature in the abnormal environmental data, and after k extractions, obtain k values of the first feature as , is the value of the first feature in the environmental data for the 1st, 2nd, 3rd... kth occurrence of an anomaly, where k is a positive integer; using the point-plotting method to draw a curve graph with the value of the first feature in the abnormal environmental data as the independent variable and the values of other types of data as the variable, ; m - 1 curves are drawn in the graph; S502. Analyze the drawn curve graph, and use the piecewise method to calculate the function for the m - 1 curves in the curve graph. The function corresponding to each curve is , is the function of the 1st, 2nd, 3rd... m - 1 curves, where m - 1 is a positive integer; S503. Perform derivative operations on the m - 1 functions obtained above to get the derivative functions , is the derivative function of the 1st, 2nd, 3rd... m - 1 curve functions, where m - 1 is a positive integer; S504. According to the above method, collect the environmental data without anomalies in the monitoring area, classify them, and generate the corresponding n types of data respectively as , draw a curve graph, calculate the curve function, and finally obtain the derivative function of each type of data in the environmental data without anomalies as , is the derivative function of the 1st, 2nd, 3rd... n - 1 curve functions, where n - 1 is a positive integer; S505. Analyze the two sets of derivative functions obtained above, and respectively extract the functions in which there is only one positive or negative situation for the values of the derivative functions. Specifically: in the abnormal environmental data, only extract the function curves where ≥0 or ≤0 in the whole function. In the environmental data without anomalies, only extract the function curves where ≥0 or ≤0 in the whole function, , ; S506. Compare the two sets of functions after extraction, and select the data type x corresponding to the variable values in the curves with the same function as the second feature in the monitoring area.

[0013] By screening the curves in the curve graph, when there is only one change situation for the derivative function of the curve, it proves that this data may have a certain linear relationship with the first feature. By comparing the curve graphs of the abnormal and non - abnormal situations in the monitoring area, finding the curves with the same change among the screened curves can locate the data types that can linearly affect the first feature in the monitoring area, and this type of data can be located as the second feature in the monitoring area.

[0014] S600 calculates the linear features of the first feature and the second feature, and realizes the real-time optimization of the hierarchical warning according to the calculated linear features of the first feature and the second feature.

[0015] Further, the calculation of the linear features of the first feature and the second feature is specifically as follows: According to the above S500, in the process of calculating the second feature, the inverse function of the curve function corresponding to the selected second feature is defined as the linear relationship between the first feature and the second feature. Let the linear relationship function between the first feature and the second feature be , in the function is the value of the first feature, x is the value of the second feature, is the linear relationship between the first feature and the second feature; The specific process of realizing the real-time optimization of the warning classification is as follows: Use a variety of sensors to collect data of the second feature in the monitoring area, and obtain the real-time value x_s of the second feature in the monitoring area. According to the linear relationship function , substitute the value x_s of the second feature as the independent variable into the function to calculate the theoretical minimum warning value _b of the first feature. Compare the calculated theoretical minimum warning value of the first feature with the actual minimum warning value _min, calculate the difference _c = _b _min. Replace the minimum warning value _min with the theoretical minimum warning value _b, and add the grading data of the warning and the difference _c to perform an addition operation to obtain the new grading data of the warning.

[0016] When warning in the monitoring area, there are other data outside the judgment standard first feature that affect the warning standard, resulting in the inability to timely warn of abnormal situations in the monitoring area. For example, when warning of a fire during mine transportation, let the temperature be the first feature and be used as the warning standard, but the humidity in the air will still affect the temperature standard for a fire.

[0017] The mine transportation safety adaptive optimization system based on video monitoring, the adaptive optimization system includes an environment perception module, a feature extraction module, a grading warning module and an intelligent optimization module; The environment perception module uses a variety of sensors to collect environmental data in the monitoring area; The feature extraction module is used to extract the first feature and the second feature according to the historical environmental data without anomalies and the environmental data with anomalies in the monitoring area; The hierarchical early warning module is used to monitor the environmental data in the monitoring area in real time, calculate the hierarchical data of the early warning according to the first feature, and then implement hierarchical early warning for abnormal environments occurring in the monitoring area according to the first feature; The intelligent optimization module is used to modify the hierarchical data in the hierarchical early warning module according to the extracted second feature.

[0018] The feature extraction module includes a first feature extraction unit and a second feature extraction unit; The first feature extraction unit is used to collect the data of abnormal environmental data and non-abnormal environmental data in the monitoring area, compare the two sets of collected data to obtain difference data, after k comparisons, count the k times of difference data, and extract the data type with the most repeated occurrences in the k times of difference data , and set as the first feature in the monitoring area; The second feature extraction unit is used to collect the data of abnormal environmental data and non-abnormal environmental data in the monitoring area, respectively draw data curve graphs for the two sets of collected data, calculate and analyze the two curve graphs to obtain, for each curve, the function and the derivative function, respectively extract the functions whose derivative function values have only one positive or negative situation, compare the two sets of functions after extraction, and select the data type x corresponding to the variable value in the curves with the same function as the second feature in the monitoring area.

[0019] The hierarchical early warning module analyzes the speed at which the number of types of environmental data that change due to the change of the first feature increases when an abnormality occurs in the monitoring area, and obtains the hierarchical data of the early warning when an abnormality occurs in the monitoring area.

[0020] The intelligent optimization module includes a feature relationship calculation unit and a real-time optimization unit; The feature relationship calculation unit is used to define the inverse function of the curve function corresponding to the second feature selected during the calculation of the second feature as the linear relationship between the first feature and the second feature, and set the linear relationship function between the first feature and the second feature as , where in the function is the value of the first feature, x is the value of the second feature, is the linear relationship between the first feature and the second feature; The real-time optimization unit is used to collect the data of the second feature in the monitoring area using multiple sensors, and according to the calculated linear relationship function between the first feature and the second feature as , modify the minimum level _min of the calculated early warning and the hierarchical data of the early warning.

[0021] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: 1. By judging the change of the first characteristic value and monitoring the change of the remaining environmental data in the monitoring area, the greater the derivative, the faster the speed of the impact caused by the abnormality in the monitoring area. Therefore, setting the warning level according to the size of the derivative can effectively classify the speed of the impact of the abnormality in the monitoring area on the environment and allocate resources reasonably.

[0022] 2. By judging the relationship between all the data in the monitoring area and the first characteristic, the second characteristic data that can linearly affect the first characteristic is obtained, and the warning classification data is optimized and modified through the second characteristic data; it can ensure that when the monitoring area is monitored in real time, the occurrence of abnormalities and the impact of abnormalities on the environment of the monitoring area are avoided. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, and do not constitute a limitation to the present invention. In the drawings: Figure 1 is a module distribution diagram of the mine transportation safety adaptive optimization system based on video monitoring of the present invention; Figure 2 is a step schematic diagram of the mine transportation safety adaptive optimization method based on video monitoring of the present invention; Figure 3 is an abnormal function curve graph of the mine transportation safety adaptive optimization method based on video monitoring of the present invention; Figure 4 is a non-abnormal function curve graph of the mine transportation safety adaptive optimization method based on video monitoring of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0025] Please refer to Figures 1-4 , the present invention provides the following technical solutions: A mine transportation safety adaptive optimization method based on video monitoring, the method comprising the following steps: S100. Use a variety of sensors to sense the environment in the monitoring area to obtain multi-source environmental data in the monitoring area; Further, the sensors adopt vision sensors, temperature and humidity sensors, and microphone sensors, and the environmental multi-source data includes temperature and humidity data, image data, and sound data in the monitoring area.

[0026] S200. Process and analyze the environmental data sensed by the sensors and perform feature extraction to obtain the first feature of the monitoring area. Further, the specific steps for performing feature extraction on the environment in the monitoring area to obtain the first feature of the monitoring area are as follows: S201. Collect the environmental data without abnormalities in the monitoring area using multiple sensors and classify them. According to the different sensors, the data collected by different sensors are divided into different categories, generating corresponding n types of data, which are respectively , respectively the 1st, 2nd, 3rd... nth types of data, where n is a positive integer; an abnormality in the monitoring area means an explosive change in multiple types of environmental data in the monitoring area.

[0027] S202. Record the environmental data with abnormalities in the monitoring area, classify the data, and according to the different sensors, divide the data collected by different sensors into different categories, generating m types of abnormal environmental data, which are respectively , respectively the 1st, 2nd, 3rd... mth types of abnormal environmental data, where m is a positive integer. S203. After classifying and collecting the environmental data when there are no abnormalities and when there are abnormalities in the monitoring area, compare the data in the two cases, and use the environmental data without abnormalities in the monitoring area to screen the environmental data with abnormalities in the monitoring area , and delete the values in the environmental data with abnormalities in the monitoring area that are the same as the environmental data without abnormalities . Designate the abnormal environmental data after deletion as the difference data , which are the 1st, 2nd, 3rd... jth types of difference data, where j is a positive integer. S204. According to the above steps, extract the environmental data with abnormalities in the monitoring area k times, and after comparing it with the environmental data without abnormalities in the monitoring area k times, generate k times of difference data. Statistically analyze the k times of difference data, and extract the data type with the most repeated occurrences in the k times of difference data , where the subscript y ranges from 1 to j, and set as the first feature in the monitoring area. The first feature generated above reflects the most significant change characteristics in the environmental data when an anomaly occurs in the monitored area. Whether an anomaly occurs in the monitored area can be judged by monitoring the first feature.

[0028] S300. Analyze the environment in the monitored area, calculate the abnormal environment index in the monitored area, and realize the hierarchical early warning of anomalies in the monitored area; Furthermore, the steps to realize the hierarchical early warning in the monitored area are as follows: S301. According to the collection of abnormal environmental data in the monitored area, after k times of comparison between the environmental data without anomalies and the environmental data with anomalies in the monitored area in S200, the number of difference data generated k times is , which are the numbers of difference data generated by comparing the environmental data without anomalies and the environmental data with anomalies in the monitored area for the 1st, 2nd, 3rd... kth times. k is a positive integer; S302. Sort the values of the first feature in ascending order, and use the point plotting method to draw a line chart of the first feature values and the number of difference data , where p ranges from 1 to k; S303. Normalize and fit the line chart to form a smooth function curve. Using the segmentation method, calculate the functional relationship of each segment in the function curve to generate a function set as , which are the functions of the 1st, 2nd, 3rd... h segments in the function curve. h is a positive integer; S304. Perform derivative operations on the piecewise functions respectively to obtain the derivative functions , which are the derivative functions of the 1st, 2nd, 3rd... h segments of the function curve. h is a positive integer; The derivative function represents the growth rate of each segment of the function curve; S305. Screen the derivative functions calculated above, select the derivative functions with values greater than 0 and sort them in ascending order. According to the sorting of the derivative functions, obtain the sorting of the corresponding independent variables as , which are the independent variables corresponding to the 1st, 2nd, 3rd... a derivative functions greater than 0 sorted in ascending order of the derivative function values , where a is a positive integer; Define the calculated as the hierarchical data for early warning; The larger the value of the derivative function of the curve function, the more it proves that at the independent variable When it increases, the variable changes faster; it can reflect that when an abnormality occurs in the monitoring area, by judging the change of the first characteristic value and the change of the remaining environmental data in the monitoring area, the greater the derivative, the faster the speed of the impact caused by the abnormality in the monitoring area. Therefore, setting the warning level according to the size of the derivative can effectively classify the speed of the impact of the abnormality in the monitoring area on the environment and allocate resources reasonably.

[0029] S306. Classify and warn the monitoring area according to the calculated classification data Perform hierarchical warning on the monitoring area and select the independent variable in the line chart The initial value is the minimum level of the warning _min, and then according to Increase the warning level in sequence.

[0030] The above-mentioned classification data for warning in the monitoring area is calculated according to the derivative. After an abnormality occurs in the monitoring area, the speed of increase in the number of changed environmental data caused by the change of the first characteristic will also be different, and the value of the derivative represents the increase speed of the changed environmental data in the monitoring area. Classify and warn the abnormality according to the different increase speeds. For example: when monitoring whether a fire may occur during mine transportation, in the monitoring area, set the temperature as the first characteristic. When the temperature is too high, it is judged that a fire abnormality occurs, and after the fire occurs, the remaining data in the monitoring area will change, and the increase speed of the number of changed data proves the size of the fire from the side. Therefore, hierarchical warning of abnormalities can be realized.

[0031] S400. Use the first characteristic obtained by calculation in the monitoring area for machine learning to construct an anomaly warning model; Furthermore, specifically constructing the anomaly warning model is as follows: Use the calculated classification data Perform machine learning using the convolutional neural network algorithm to construct a hierarchical warning model; use sensors to perceive the environment in the monitoring area in real time, and extract the first characteristic value _s in the environmental data of the monitoring area. Input the extracted first characteristic value _s into the constructed hierarchical warning model, and judge that when the extracted first characteristic value _s > _min, the environment in the monitoring area is abnormal. Then compare _s with the classification data in sequence. When it is judged that _s > , issue a warning of the corresponding level to the monitoring area, where i belongs to 1 - a.

[0032] S500. Analyze the environment during the early warning, and extract the second feature of the monitoring area; Further, the specific steps for extracting the second feature of the monitoring area are as follows: S501. Collect the data of the abnormal environment in the monitoring area, extract the environmental data of the abnormal occurrences in the monitoring area k times according to the method described in S200, classify the environmental data extracted each time, and generate m types of abnormal environmental data, respectively , extract the first feature from the abnormal environmental data, and obtain the values of k first features after k extractions as , are the values of the first feature in the environmental data of the abnormal occurrences at the 1st, 2nd, 3rd... kth times, and k is a positive integer; Use the point plotting method to draw a curve graph with the value of the first feature in the abnormal environmental data as the independent variable and the values of other types of data as the variable, ; m - 1 curves are drawn in the graph; S502. Analyze the drawn curve graph, calculate the functions for the m - 1 curves in the curve graph using the segmentation method, and the function corresponding to each curve is , are the functions of the 1st, 2nd, 3rd... m - 1 curves, and m - 1 is a positive integer; S503. Perform derivative operations on the m - 1 functions calculated above to obtain the derivative functions , are the derivative functions of the functions of the 1st, 2nd, 3rd... m - 1 curves, and m - 1 is a positive integer; S504. According to the above method, collect the environmental data that has not had an abnormal occurrence in the monitoring area, classify it, and generate the corresponding n types of data, respectively , draw a curve graph, calculate the curve function, and finally obtain the derivative function of each type of data in the environmental data without abnormal occurrence as , are the derivative functions of the functions of the 1st, 2nd, 3rd... n - 1 curves, and n - 1 is a positive integer; S505. Analyze the two sets of derivative functions calculated above, and respectively extract the functions whose derivative function values only have one positive or negative situation, specifically: in the abnormal environmental data, only extract the function curves where ≥0 or ≤0 in the whole function, and in the environmental data without abnormal occurrence, only extract the function curves where ≥0 or ≤0 in the whole function, , ; S506. Compare the two sets of functions after extraction, and select the data type x corresponding to the variable values in the curves with the same functions as the second feature within the monitoring area.

[0033] By screening the curves in the curve graph, when there is only one change situation in the derivative function of the curve, it proves that this data may have a certain linear relationship with the first feature. By comparing the curve graphs with and without anomalies within the monitoring area, the data types that can linearly affect the first feature can be located by finding the curves with the same changes among the screened curves. Such data can be positioned as the second feature within the monitoring area.

[0034] S600. Calculate the linear features of the first feature and the second feature, and realize the real-time optimization of the hierarchical early warning according to the calculated linear features of the first feature and the second feature.

[0035] Furthermore, the specific calculation of the linear features of the first feature and the second feature is as follows: According to the above S500, during the process of calculating the second feature, the inverse function of the curve function corresponding to the selected second feature is defined as the linear relationship between the first feature and the second feature. Let the linear relationship function between the first feature and the second feature be , in the function is the value of the first feature, x is the value of the second feature, is the linear relationship between the first feature and the second feature; The specific process of realizing the real-time optimization of the warning classification is as follows: Use a variety of sensors to collect data on the second feature within the monitoring area to obtain the real-time value x_s of the second feature within the monitoring area. According to the linear relationship function , substitute the value x_s of the second feature as the independent variable into the function to calculate the theoretical minimum warning value _b of the first feature. Compare the calculated theoretical minimum warning value of the first feature with the actual minimum warning value _min, calculate the difference _c = _b _min. Replace the minimum warning value _min with the theoretical minimum warning value _b, and add the grading data of the warning and the difference _c to perform an addition operation to obtain the new grading data of the warning.

[0036] When giving an early warning within the monitoring area, there are other data outside the first feature of the judgment standard that affect the early warning standard, resulting in the inability to give an early warning of abnormal situations within the monitoring area in a timely manner. For example, when giving an early warning of a fire during mine transportation, the temperature is set as the first feature and used as the early warning standard, but the humidity in the air will still affect the temperature standard for a fire.

[0037] An adaptive optimization system for mine transportation safety based on video monitoring, the adaptive optimization system includes an environmental perception module, a feature extraction module, a hierarchical early warning module, and an intelligent optimization module; The environmental perception module uses a variety of sensors to collect environmental data within the monitoring area; The feature extraction module is used to extract the first feature and the second feature according to the environmental data without abnormalities and with abnormalities in the historical monitoring area; The hierarchical early warning module is used to monitor the environmental data within the monitoring area in real time, calculate the hierarchical data of the early warning according to the first feature, and then give a hierarchical early warning when an abnormal environment occurs within the monitoring area according to the first feature; The intelligent optimization module is used to modify the hierarchical data in the hierarchical early warning module according to the extracted second feature.

[0038] The feature extraction module includes a first feature extraction unit and a second feature extraction unit; The first feature extraction unit is used to collect the environmental data with abnormalities and without abnormalities in the monitoring area, compare the two sets of collected data to obtain the difference data. After k comparisons, count the k times of difference data, and extract the data type with the most repeated occurrences in the k times of difference data , and set it as the first feature within the monitoring area; The second feature extraction unit is used to collect the environmental data with abnormalities and without abnormalities in the monitoring area, respectively draw data curves for the two sets of collected data, calculate and analyze the two curves, and obtain, for each curve, the function and the derivative function. Respectively extract the functions whose derivative values have only one positive or negative situation, compare the two sets of functions after extraction, and select the data type x corresponding to the variable values in the curves with the same function as the second feature within the monitoring area.

[0039] The hierarchical early warning module analyzes the speed at which the number of types of environmental data that change due to the change of the first feature increases when an abnormality occurs within the monitoring area, and obtains the hierarchical data of the early warning when an abnormality occurs within the monitoring area.

[0040] The intelligent optimization module includes a feature relationship calculation unit and a real-time optimization unit; The feature relationship calculation unit is used to define the inverse function of the curve function corresponding to the second feature selected during the process of calculating the second feature as the linear relationship between the first feature and the second feature. Let the linear relationship function between the first feature and the second feature be , where in the function is the value of the first feature, x is the value of the second feature, is the linear relationship between the first feature and the second feature; The real-time optimization unit is used to collect data of the second feature in the monitoring area by using multiple sensors, and according to the calculated linear relationship function between the first feature and the second feature as , modify the calculated minimum level _min of the warning and the warning classification data.

[0041] Example 1 First, monitor a certain mine transportation working environment, and after calculation, the first feature data and warning classification data in the monitoring area have been obtained. Let the first feature data be , and the warning classification data be Now, extract the second feature in the monitoring area, and collect the environmental data where anomalies occur and do not occur in the monitoring area as shown in Table 1 and Table 2 below:

[0042] Table 1

[0043] Table 2 Draw a curve graph according to the content in the above table, as Figure 3 and Figure 4 stated. In the figure, Series 1 represents the curve of and and, and Series 2 represents the curve of and and; Through the analysis and calculation of the curves in the two figures, the function in the anomaly-occurring curve graph is obtained as , and the function in the non-anomaly-occurring curve graph is , . Respectively perform derivative operations on the four functions to obtain the corresponding derivative functions as , ; By calculating the derivative functions of the curve functions in the two figures respectively, it can be easily obtained that in the anomaly-occurring curve graph there is only one value, and in the non-anomaly-occurring curve graph there is only one value; then compare the two functions obtained by screening, and Both are less than 0, so it is determined that the second feature in the monitoring area is data and the data types represented by

[0044] It should be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device.

[0045] Finally, it should be noted that the above are only preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. The adaptive optimization method for mine transportation safety based on video surveillance is characterized by: The method comprises the following steps: S100, using a variety of sensors to sense the environment in the monitoring area, and obtaining multi-source environmental data in the monitoring area; S200, processing and analyzing the environmental data sensed by the sensor and performing feature extraction to obtain a first feature of the monitored area; S300, analyzing the environmental data in the monitoring area, calculating the change speed of abnormal environmental data in the monitoring area, and implementing graded warning of abnormalities in the monitoring area; S400, performing machine learning using the first feature obtained by calculation in the monitoring area to build an abnormal warning model; S500, analyzing the environment during the warning, and extracting the second feature of the monitoring area; S600: Calculate and obtain linear features of the first feature and the second feature, and implement real-time optimization of graded warning according to the calculated linear features of the first feature and the second feature.

2. The method for adaptive optimization of mine transportation safety based on video monitoring according to claim 1 is characterized in that: The specific steps of extracting the features of the environment in the monitoring area to obtain the first feature of the monitoring area in S200 are: S201, using a variety of sensors to collect and classify environmental data without abnormalities in the monitoring area, and according to the different sensors, the data collected by different sensors are divided into different types, and the corresponding n types of data are generated respectively. , They are the 1st, 2nd, 3rd...nth types of data respectively, and n is a positive integer; S202, record the abnormal environmental data in the monitoring area, classify the data, and divide the data collected by different sensors into different types according to the different sensors, and generate m types of abnormal environmental data respectively. , They are the 1st, 2nd, 3rd...mth abnormal environmental data, where m is a positive integer; S203, after classifying and collecting the environmental data when no abnormality occurs and when an abnormality occurs in the monitoring area, the data in the two situations are compared, and the environmental data when no abnormality occurs in the monitoring area is used to compare the environmental data when no abnormality occurs. Abnormal environmental data in the monitoring area Screen and monitor abnormal environmental data in the area Environmental data with no abnormality The same value is deleted, and the abnormal environment data after deletion is defined as difference data , is the 1st, 2nd, 3rd, ...jth difference data, where j is a positive integer; S204: According to the above steps, extract the environmental data with abnormalities in the monitoring area for k times, and then compare them with the environmental data without abnormalities in the monitoring area for k times to generate k difference data, perform statistics on the k difference data, and extract the data type with the most repetition times in the k difference data. , the subscript y takes values ​​from 1 to j, Set as the first feature in the monitoring area.

3. The method for adaptive optimization of mine transportation safety based on video monitoring according to claim 1 is characterized in that: The steps of implementing the graded warning in the monitoring area in S300 are: S301, based on the collection of abnormal environment data in the monitoring area, after performing k comparisons between the data of no abnormal environment and the data of abnormal environment in the monitoring area in S200, the number of generated k difference data is respectively , is the number of difference data generated by comparing the normal environmental data and the abnormal environmental data in the monitoring area for the 1st, 2nd, 3rd...kth time, where k is a positive integer; S302, the first feature is sorted in ascending order Sort the values ​​and draw the first feature using the point drawing method Number of numerical and differential data Line graph, p ranges from 1 to k; S303, normalize and fit the line graph to form a smooth function curve, use the segmentation method to calculate the functional relationship of each segment in the function curve, and generate a function set as , is the function of the 1st, 2nd, 3rd...hth segments in the function curve, where h is a positive integer; S304, perform derivative operations on the piecewise functions respectively to obtain the derivative function of each section of the function curve , is the derivative function of the 1st, 2nd, 3rd...hth segment function curve, h is a positive integer; the derivative function describes the growth rate of each segment function curve; S305: Screen the derivative functions calculated above, select the derivative functions whose values ​​are greater than 0 and sort them in ascending order, and obtain the corresponding independent variables according to the order of the derivative functions. The sorting is , is the independent variable corresponding to the first, second, third, ... a derivative functions greater than 0 in ascending order of derivative function value , a is a positive integer; the calculated Classification data for early warning; S306, obtaining classification data according to calculation Conduct graded warnings in the monitoring area and select independent variables in the line graph The initial value is the minimum warning level _min, then according to The warning levels increase in order.

4. The method for adaptive optimization of mine transportation safety based on video monitoring according to claim 1 is characterized in that: The abnormal warning model is constructed in S400 specifically as follows: The calculated classification data Use convolutional neural network algorithm for machine learning to build a hierarchical warning model; use sensors to perceive the environment in the monitoring area in real time and extract the first characteristic value in the environmental data in the monitoring area _s, the first feature value extracted _s is input into the constructed hierarchical warning model to determine when the first feature value extracted _s> _min, the environment in the monitoring area is abnormal, and then _s and hierarchical data Compare them one by one, and when judging _s> When , an early warning of the corresponding level is issued to the monitored area, and i belongs to 1-a.

5. The method for adaptive optimization of mine transportation safety based on video monitoring according to claim 1 is characterized in that: The specific steps of extracting the second feature of the monitoring area in S500 are: S501, collect abnormal environmental data in the monitoring area, extract k abnormal environmental data in the monitoring area according to the method described in S200, classify the environmental data extracted each time, and generate m types of abnormal environmental data respectively. , extract the first feature in the abnormal environment data, and after k extractions, get the value of k first features , is the value of the first feature in the abnormal environmental data for the 1st, 2nd, 3rd...kth time, k is a positive integer; the first feature of the abnormal environmental data is plotted using the point drawing method The value of is an independent variable, other types of data values is a graph of the variables, ; m-1 curves are drawn in the figure; S502, analyze the drawn curve graph, and use the segmentation method to calculate the function of the m-1 curves in the curve graph. The function corresponding to each curve is , is the function of the 1st, 2nd, 3rd...m-1th curve, where m-1 is a positive integer; S503, perform derivative operation on the m-1 functions obtained by the above calculation to obtain a derivative function , is the derivative function of the 1st, 2nd, 3rd...m-1th curve function, where m-1 is a positive integer; S504: According to the above method, the environmental data without abnormality in the monitoring area is collected and classified to generate corresponding n types of data. , draw a curve graph, calculate the curve function, and finally get the derivative function of each data in the environmental data without abnormality: , is the derivative function of the 1st, 2nd, 3rd...n-1th curve function, n-1 is a positive integer; S505: Analyze the two groups of derivative functions obtained by the above calculations, and respectively extract the functions in which the derivative function has only one positive or negative value. Specifically, in the abnormal environment data, only extract the derivative function of the entire function. ≥0 or ≤0 function curve, only extract the entire function in the environment without abnormal data ≥0 or ≤0 function curve, , ; S506: Compare the two groups of functions after extraction, and select the data type x corresponding to the variable value in the curve with the same function as the second feature in the monitoring area.

6. The method for adaptive optimization of mine transportation safety based on video monitoring according to claim 1 is characterized in that: The linear features of the first feature and the second feature calculated in S600 are specifically: According to the above S500, in the process of calculating the second feature, the inverse function of the curve function corresponding to the selected second feature is defined as the linear relationship between the first feature and the second feature. The linear relationship function between the first feature and the second feature is assumed to be , in the function is the value of the first feature, x is the value of the second feature, is the linear relationship between the first feature and the second feature; The specific process of achieving real-time optimization of early warning classification is as follows: Use a variety of sensors to collect data on the second feature in the monitoring area, and obtain the real-time value x_s of the second feature in the monitoring area. According to the linear relationship function , substitute the value x_s of the second feature as the independent variable into the function to calculate the theoretical minimum warning value of the first feature _b, calculate the theoretical minimum warning value of the first feature and the actual minimum warning value _min to compare and calculate the difference _c= _b _min, the minimum warning value _min uses the theoretical minimum warning value _b is replaced and the warning classification data is Mean and Difference _c performs addition operation to obtain new warning classification data.

7. The mine transportation safety adaptive optimization system based on video surveillance is characterized by: The mine transportation safety adaptive optimization system includes an environmental perception module, a feature extraction module, a hierarchical warning module, and an intelligent optimization module; The environmental perception module uses a variety of sensors to collect environmental data in the monitoring area; The feature extraction module is used to extract the first feature and the second feature based on the data of no abnormal environment and the data of abnormal environment in the historical monitoring area; The hierarchical warning module is used to monitor the environmental data in the monitoring area in real time, calculate the hierarchical data of the warning according to the first feature, and then implement hierarchical warning when an abnormal environment occurs in the monitoring area according to the first feature; The intelligent optimization module is used to modify the classification data in the classification warning module according to the extracted second feature.

8. The mine transportation safety adaptive optimization system based on video monitoring according to claim 7 is characterized by: The feature extraction module includes a first feature extraction unit and a second feature extraction unit; The first feature extraction unit is used to collect data on abnormal environments and data on non-abnormal environments in the monitoring area, compare the two sets of collected data to obtain difference data, and after k comparisons, perform statistics on the k times of difference data to extract the data type with the most repetition times in the k times of difference data. ,Will Set as the first feature in the monitoring area; The second feature extraction unit is used to collect data on abnormal environments and non-abnormal environments in the monitoring area, draw data curve graphs for the two sets of collected data, calculate and analyze the two curve graphs to obtain the function and derivative of each curve, extract the functions in which the derivative value has only one positive and negative situation, compare the two sets of functions after extraction, and select the data type x corresponding to the variable value in the curve with the same function as the second feature in the monitoring area.

9. The mine transportation safety adaptive optimization system based on video monitoring according to claim 7 is characterized by: The hierarchical warning module obtains hierarchical data for warning when an abnormality occurs in the monitoring area by analyzing the speed at which the number of other types of environmental data that have changed increases due to a change in the first feature when an abnormality occurs in the monitoring area.

10. The mine transportation safety adaptive optimization system based on video monitoring according to claim 7 is characterized in that: The intelligent optimization module includes a feature relationship calculation unit and a real-time optimization unit; The feature relationship calculation unit is used to define the inverse function of the curve function corresponding to the second feature selected in the process of calculating the second feature as the linear relationship between the first feature and the second feature, and the linear relationship function between the first feature and the second feature is set as , in the function is the value of the first feature, x is the value of the second feature, is the linear relationship between the first feature and the second feature; The real-time optimization unit is used to collect data of the second feature in the monitoring area using multiple sensors, and the linear relationship function between the first feature and the second feature obtained by calculation is: , the minimum level of the calculated warning _min and warning classification data are modified.

Images