Automatic monitoring system of ground film based on multimodal data fusion
Through the mulch automation monitoring system with multimodal data fusion, the problem of low degree of mulch monitoring in the existing technology is solved, more accurate and comprehensive data analysis and management is achieved, and management efficiency and automation level are improved.
Patent Information
- Application Number
- CN202410342689.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-25
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-03-25
Smart Images

Figure CN118293973B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of ground film monitoring and management, and in particular to a ground film automatic monitoring system based on multi-modal data fusion. Background Art
[0002] Multimodal data fusion refers to combining multiple types of data from different sensors or data sources to obtain more comprehensive and accurate information. These data types can include text, images, audio, and video. The purpose of fusing these multimodal data is to improve the understanding and analysis capabilities of complex phenomena or events. By fusing information from different data sources, the performance and efficiency of the system can be improved, and more diverse and comprehensive analysis and applications can be achieved.
[0003] Ground film is a thin film material used in agricultural production, usually made of polyethylene or other plastic materials. It is laid on the soil surface and covers the bottom of the crops. It plays the role of covering, keeping warm, retaining moisture, and inhibiting weed growth.
[0004] The existing methods of monitoring and managing ground film have too low a degree of automation and rely too much on manual monitoring and management of ground film. Only by accurate data analysis of the monitored ground film area can efficient management of the ground film be achieved. The data collected in the ground film area is an important basis for data analysis. How to obtain more comprehensive and accurate data and perform automated management of the corresponding ground film area based on the results of data analysis is an issue that we urgently need to consider. Summary of the invention
[0005] In order to solve the above problems, the object of the present invention is to provide an automatic monitoring system for ground film based on multimodal data fusion.
[0006] The object of the present invention can be achieved through the following technical solutions: a ground film automatic monitoring system based on multimodal data fusion, comprising a monitoring center, wherein the monitoring center is communicatively connected with a data acquisition module, a data processing and fusion module, a data analysis module, an alarm prompt module and a ground film automatic management module;
[0007] The data acquisition module includes an equipment deployment unit and a multi-source data acquisition unit; the equipment deployment unit is used to deploy ground film related working equipment and ground film related acquisition equipment; the multi-source data acquisition unit is used to start the data acquisition work of the ground film related acquisition equipment, and then obtain different types of ground film related monitoring data;
[0008] The data processing and fusion module is used to fuse different types of ground film related monitoring data, thereby generating ground film comprehensive monitoring data, and performing modular association and multi-path dynamic transmission for the ground film comprehensive monitoring data;
[0009] The data analysis module is used to receive the comprehensive monitoring data of ground film from multi-path dynamic transmission, integrate the comprehensive monitoring data of ground film from different transmission paths, perform data analysis on the integrated comprehensive monitoring data of ground film, and then generate a data analysis report;
[0010] The alarm prompt module compares the data analysis report with the preset alarm threshold, and then determines whether to generate an alarm prompt message according to the comparison result, and determines whether to perform feedback change of the alarm threshold according to the alarm success rate of the alarm prompt message;
[0011] The mulch film automation management module is provided with a data reading unit and an automation management unit; the data reading unit is used to read the alarm prompt information and the equipment parameter information stored in the monitoring center; the automation management unit is used to execute the corresponding mulch film automation management measures according to the alarm prompt information and the equipment parameter information.
[0012] Furthermore, the process of deploying ground film related working equipment and ground film related collection equipment includes:
[0013] Select the mulch area that needs to be monitored, select and schedule the mulch-related working equipment and mulch-related collection equipment, and obtain the working coverage area of the mulch-related working equipment, recorded as S 工设 , obtain the collection coverage area of the ground film related collection equipment, recorded as S 采集 , the area of the mulch film area to be monitored is recorded as S, and according to S and S 工设 A number of working equipment areas are divided on the ground film area and numbered as i, where i = 1, 2, 3, ..., n. 采集 A number of collection equipment areas are divided and numbered as j, where j = 1, 2, 3, ..., m, where n and m are natural numbers greater than 0;
[0014] In the working equipment areas numbered i and the collecting equipment areas numbered j, ground film-related working equipment and ground film-related collecting equipment are deployed respectively, and their corresponding communication sequences are obtained, which are recorded as A[i] and B[i] respectively. The ground film-related working equipment includes automatic feeding equipment and automatic exhaust equipment, and the ground film-related collecting equipment includes sensor collection modules and monitoring cameras. The equipment parameter information of the ground film-related working equipment is obtained and transmitted to the monitoring center for storage.
[0015] Furthermore, the process of obtaining different types of ground film related monitoring data includes:
[0016] Set the communication link time and data collection period, and record the data collection period as T 采集 , there is T 采集=[t1, t2], where t1 is the start time of the data collection period, and t2 is the end time of the data collection period. At the communication link time, the communication sequences A[i] and B[i] are obtained, and then the data communication with the film-related working equipment corresponding to A[i] and the film-related collection equipment corresponding to B[i] is established. At the time point corresponding to t1, the data collection work of the film-related collection equipment is started, and then different types of film-related monitoring data are obtained. The types of film-related monitoring data include sensor data and image frame data;
[0017] Among them, the sensor data is collected and acquired by the sensor acquisition module. The sensor data includes soil moisture, soil temperature, soil pH and external illumination. The image frame data includes single-frame images and multi-frame images. The image frame data is collected by the monitoring camera. At the corresponding time t2, the monitoring data related to different types of ground films are collected. The number j corresponding to the current collection device area is bound to the ground film related monitoring data, and then the ground film related monitoring data of several collection device areas are obtained.
[0018] Furthermore, the process of generating comprehensive monitoring data of ground film includes:
[0019] Different types of ground film related monitoring data are used as data objects for data fusion, and the data format of the data objects is converted into a preset data fusion standard format. The sensor data and image frame data after the conversion are used as fusion element one and fusion element two, respectively, and are marked as Rh1 and Rh2, respectively. Before data fusion of Rh1 and Rh2, different types of feature parameters are extracted for Rh1 and Rh2, respectively. The types of feature parameter extraction include feature curve segmentation marking and feature image hierarchical annotation. Feature curve segmentation marking is performed on Rh1, and feature image hierarchical annotation is performed on Rh2.
[0020] Generate characteristic curves and reasonable value intervals through segmented marking of characteristic curves, mark sensor data in the reasonable value interval of the characteristic curve as cold monitoring data of fusion element one, mark sensor data not in the reasonable value interval as hot monitoring data of fusion element one, and obtain image feature data of fusion element two through hierarchical annotation of characteristic images;
[0021] Set the data text, enter the cold monitoring data and hot monitoring data corresponding to fusion element one into the data text, enter the image feature data corresponding to fusion element two into the data text, and then generate different text fields corresponding to fusion element one and fusion element two, associate the same parts of the text fields in fusion element one and fusion element two, and use them as a text association item. Do not perform any operation on the different parts of the text fields in fusion element one and fusion element two, perform data fusion on several text association items, and then generate comprehensive monitoring data for ground film through data fusion.
[0022] Furthermore, the process of modularizing and associating the comprehensive monitoring data of ground film and dynamically transmitting the data in multiple paths includes:
[0023] Set up several module nodes, each module node corresponds to a data transmission protocol, and number the several module nodes and record them as k, k = 1, 2, 3, ..., q, where q is a natural number greater than 0. The amount of data of the comprehensive monitoring data of the ground film is obtained, which is recorded as C 传输 , distribute the comprehensive monitoring data of ground film to several module nodes, and the amount of comprehensive monitoring data of ground film distributed to the module node numbered k is C[k] 分配 , the transmission capacity corresponding to the module node numbered k is recorded as C[k] 承载 , we have C[k] 分配 =C[k] 承载 When each module node receives the allocated comprehensive monitoring data of ground film, the modular association of all module nodes is successful. Otherwise, the comprehensive monitoring data of ground film will continue to be allocated to the module nodes that have not received it.
[0024] Read the data transmission protocol corresponding to the module node numbered k, and then establish a transmission path between the current module node numbered k and the data analysis module, repeat the above operation, and then establish several transmission paths, each of the transmission paths is allowed to replace the data transmission of the corresponding part of the ground film comprehensive monitoring data of the adjacent transmission path under the set dynamic transmission conditions, the ground film comprehensive monitoring data transmitted through each transmission path is marked as segmented transmission data, and the dynamic transmission conditions are set as follows: if the real-time packet loss rate, transmission signal strength and delay rate of the current transmission path are all lower than their respective preset qualified thresholds, the segmented transmission data of the current transmission path will be transferred to the adjacent transmission path, and the data will be transmitted by the adjacent transmission path.
[0025] Furthermore, the process of integrating the comprehensive monitoring data of ground film of different transmission paths includes:
[0026] A data integration program is set, and a data receiving period and a data integration period corresponding to the data integration program are set, the ground film comprehensive monitoring data from multi-path dynamic transmission is received during the data receiving period, and the segmented transmission data corresponding to different transmission paths are integrated during the data integration period;
[0027] During the data receiving period, an encrypted receiving space is set up, and the segmented transmission data corresponding to different transmission paths are received into the encrypted receiving space, and the encrypted receiving space is encrypted to generate a corresponding space key, and the decryption key set corresponding to the space key is obtained, and then the segmented transmission data in the encrypted receiving space is decrypted and obtained by the decryption key. A corresponding data integration node is generated according to each transmission path, and the segmented transmission data corresponding to the transmission path is stored in the data integration node. Each data integration node is provided with a corresponding integration point, and the corresponding integration points of every two data integration nodes are connected, and then all the data integration nodes are connected. When all the data integration nodes are connected successfully, all the segmented transmission data are synchronously integrated into comprehensive monitoring data of ground film.
[0028] Furthermore, the process of analyzing the integrated ground film comprehensive monitoring data and generating data analysis reports includes:
[0029] Obtain the integrated comprehensive monitoring data of ground film, and import it into the set data analysis program for data analysis. The data analysis program is pre-entered with different types of report field sets, and each type of report field set is associated with a section of ground film analysis data content;
[0030] The bag-of-words model is set up to perform word-class grouping and word frequency statistics on the comprehensive monitoring data of ground film. The integrated comprehensive monitoring data of ground film is input into the bag-of-words model as model data. The bag-of-words model performs word-class grouping according to the model data, and then generates several different monitoring data vocabulary groups by grouping. The monitoring data vocabulary group includes several monitoring data vocabulary. The word frequency of each monitoring data vocabulary is obtained through word frequency statistics, which is recorded as f. The abnormal word frequency interval is set and recorded as Ω. 词频 , if f∈Ω 词频 , then mark the word frequency as abnormal word frequency, and remove the monitoring data vocabulary corresponding to the abnormal word frequency. If f Ω 词频 , it means that the word frequency is normal and no operation is performed;
[0031] The monitoring data vocabulary included in the monitoring data vocabulary group after the elimination operation is marked as a monitoring field. The report field set includes several report fields. The monitoring data vocabulary group and the report field set are matched to obtain a matching rate. The matching rate is recorded as R, R = (Z 匹 / Z 总 )×100%, where Z 匹 Indicates the number of successful matches between the monitoring fields in the monitoring data vocabulary group and the report fields in the report field set, Z 总 Indicates the total number of report fields in the report field set, sets the matching achievement rate, denoted as R`;
[0032] When R≥R`, a section of mulch analysis data content corresponding to the current report field set is used as the data to be filled in the report. When R<R`, no operation is performed, and all the report data to be filled that meet the condition R≥R` are recorded in the set blank report text, and then the blank report text is converted into a data analysis report.
[0033] Furthermore, the process of comparing the data analysis report with the preset alarm threshold and then determining whether to generate an alarm prompt information according to the comparison result includes:
[0034] The data analysis report includes a number of report items, each of which is provided with a corresponding project indicator. The number of report items is numbered and recorded as e, where e=1, 2, 3, ..., v, where v is a natural number greater than 0. The project indicator corresponding to the report item numbered e is obtained from the data analysis report, and then the indicator risk measurement coefficient corresponding to the project indicator is obtained and recorded as γ[e]. A corresponding alarm threshold is set for the project indicator corresponding to the report item numbered e and recorded as τ[e];
[0035] Compare the numerical values of γ[e] and τ[e], and generate different comparison results, and then decide whether to generate an alarm prompt information based on the comparison results. The comparison results include result one and result two. When γ[e]≥τ[e], the generated comparison result is result one, and the content of result one is: the numerical value of the project indicator corresponding to the current report item is initially analyzed to be abnormal; when γ[e]<τ[e], the generated comparison result is result two, and the content of result two is: the numerical value of the project indicator corresponding to the current report item is initially analyzed to be within the normal range; when the comparison result is result one, an alarm prompt information is generated, and when the comparison result is result two, no alarm prompt information is generated.
[0036] Further, the process of determining whether to perform feedback change of the alarm threshold according to the alarm success rate of the alarm prompt information includes:
[0037] Get the alarm success rate of the alarm prompt information, recorded as Sc, set the feedback change interval of the alarm threshold, and record it as Ω 反馈 , there is Ω 反馈 =[0%,u], where u is set to 70% and can be changed;
[0038] When Sc∈Ω 反馈 When , the alarm threshold is changed;
[0039] When Sc Ω 反馈 , no feedback change of the alarm threshold is performed;
[0040] The feedback changes include regular changes, event-driven changes, and custom changes.
[0041] Furthermore, the process of reading the alarm prompt information and the equipment parameter information stored in the monitoring center and then executing the corresponding mulch film automation management measures includes:
[0042] The data reading unit reads the alarm prompt information, establishes data communication between the data reading unit and the monitoring center, and then obtains the device parameter information stored in the monitoring center, and transmits the alarm prompt information and the device parameter information to the automation management unit;
[0043] The equipment parameter information includes feeding automation parameter information and exhaust automation parameter information;
[0044] Execute different ground film automation management measures according to the alarm prompt information and equipment parameter information. The ground film automation management measures include fault management measures and ground film operation management measures. Among them, the fault management measures are generated by the alarm prompt information, and the ground film operation management measures are generated by the equipment parameter information. The ground film operation management measures include feeding measures and exhaust measures;
[0045] A solution library is set up, in which specific implementation plans of several mulch film operation management measures are stored, and the feeding automation parameter information and the exhaust automation parameter information are input into the solution library, and then the specific implementation plans of the corresponding feeding measures and exhaust measures in the solution library are matched and obtained;
[0046] The automatic management unit starts the automatic feeding equipment and automatic exhaust equipment in the monitored mulch film area according to the specific implementation plan of the feeding measures and exhaust measures, and then implements the corresponding automatic feeding and automatic exhaust operations;
[0047] The fault management measures include fault handling priorities, a list of fault-related handling tools, and a list of fault handling personnel. The relevant personnel included in the fault handling personnel list are arranged to carry the fault handling tools corresponding to the fault-related handling tool list, and perform fault handling in the monitored mulch area in order of fault handling priority from high to low.
[0048] Compared with the prior art, the present invention has the following beneficial effects:
[0049] 1. By deploying a reasonable number of ground film-related working equipment and ground film-related collection equipment in the monitored ground film area, and starting the data collection work of the ground film-related collection equipment, different types of ground film-related monitoring data are obtained, and different types of ground film-related monitoring data are data-fused to generate ground film comprehensive monitoring data. The types of ground film-related monitoring data include image data and sensor data. The data fusion of the two improves the dimension and comprehensiveness of the data to a certain extent, and modular association and multi-path dynamic transmission are performed for the ground film comprehensive monitoring data. The ground film comprehensive monitoring data from different transmission paths is received and integrated through the data analysis module, and data analysis is performed on the integrated ground film comprehensive monitoring data. Among them, modular association and multi-path dynamic transmission enhance the accuracy of the ground film comprehensive monitoring data, and accurate and comprehensive ground film comprehensive monitoring data is a solid foundation for subsequent data analysis. Therefore, the accuracy of data analysis is improved to a certain extent.
[0050] 2. Compare the data analysis report generated after data analysis with the preset alarm threshold, and decide whether to generate alarm prompt information based on the comparison result, and decide whether to make feedback changes to the alarm threshold based on the alarm success rate of the alarm prompt information; read the alarm prompt information and the equipment parameter information stored in the monitoring center through the set ground film automation management module, and then execute the corresponding ground film automation management measures. On the one hand, it improves the degree of automation of ground film monitoring and management, and effectively avoids the influence of human subjective factors. On the other hand, through the feedback of changes in the alarm threshold, the generation of alarm prompt information is more flexible and the alarm mechanism is more targeted and perfect. BRIEF DESCRIPTION OF THE DRAWINGS
[0051] Figure 1 It is a schematic diagram of the present invention. DETAILED DESCRIPTION
[0052] like Figure 1 As shown, the automatic monitoring system for ground film based on multimodal data fusion includes a monitoring center, which is communicatively connected with a data acquisition module, a data processing and fusion module, a data analysis module, an alarm prompt module and a ground film automatic management module;
[0053] The data acquisition module includes an equipment deployment unit and a multi-source data acquisition unit; the equipment deployment unit is used to deploy ground film related working equipment and ground film related acquisition equipment; the multi-source data acquisition unit is used to start the data acquisition work of the ground film related acquisition equipment, and then obtain different types of ground film related monitoring data;
[0054] The data processing and fusion module is used to fuse different types of ground film related monitoring data, thereby generating ground film comprehensive monitoring data, and performing modular association and multi-path dynamic transmission for the ground film comprehensive monitoring data;
[0055] The data analysis module is used to receive the comprehensive monitoring data of ground film from multi-path dynamic transmission, integrate the comprehensive monitoring data of ground film from different transmission paths, perform data analysis on the integrated comprehensive monitoring data of ground film, and then generate a data analysis report;
[0056] The alarm prompt module compares the data analysis report with the preset alarm threshold, and then determines whether to generate an alarm prompt message according to the comparison result, and determines whether to perform feedback change of the alarm threshold according to the alarm success rate of the alarm prompt message;
[0057] The mulch film automation management module is provided with a data reading unit and an automation management unit; the data reading unit is used to read the alarm prompt information and the equipment parameter information stored in the monitoring center; the automation management unit is used to execute the corresponding mulch film automation management measures according to the alarm prompt information and the equipment parameter information.
[0058] Specifically, the process of deploying ground film related working equipment and ground film related collection equipment through the equipment deployment unit includes:
[0059] Select the film area that needs to be monitored, select and schedule the film-related working equipment and film-related collection equipment through the equipment deployment unit, obtain the working coverage area of the film-related working equipment, and record it as S 工设 , obtain the collection coverage area of the ground film related collection equipment and record it as S 采集 ;
[0060] The area of the mulch film area to be monitored is recorded as S. According to S and S 工设 Calculate and divide the ground film area into a number of working equipment areas, number the number of working equipment areas, and record the number as i, i = 1, 2, 3, ..., n, where n is a natural number greater than 0. Similarly, according to S and S 采集 Calculation is performed to divide a number of acquisition device areas, and the number of acquisition device areas is numbered, recorded as j, where j=1, 2, 3, ..., m, where m is a natural number greater than 0;
[0061] Deploy the ground film related working equipment in the working equipment area numbered i, and obtain the communication sequence of the ground film related working equipment, which is recorded as A[i]; deploy the ground film related collection equipment in the collection equipment area numbered j, and obtain the communication sequence of the ground film related collection equipment, which is recorded as B[i];
[0062] The ground film related working equipment includes automatic feeding equipment and automatic exhaust equipment, and the ground film related collection equipment includes a sensor collection module and a monitoring camera, which obtains equipment parameter information of the ground film related working equipment in real time and transmits the equipment parameter information to the monitoring center for storage.
[0063] Specifically, the process of starting the data collection work of the ground film related collection equipment through the multi-source data collection unit and then obtaining different types of ground film related monitoring data includes:
[0064] The multi-source data acquisition unit sets the communication link time and the data acquisition period, and the data acquisition period is recorded as T 采集 , there is T 采集 =[t1, t2], where t1 is the start time of the data collection period and t2 is the end time of the data collection period;
[0065] At the communication link moment, the multi-source data acquisition unit acquires the communication sequences A[i] and B[i], and then establishes data communication with the film-related working equipment corresponding to A[i] and the film-related acquisition equipment corresponding to B[i]. When the data communication is established, the data acquisition work of the film-related acquisition equipment is started at the corresponding time point t1, and different types of film-related monitoring data are acquired through the data acquisition work, and the types of the film-related monitoring data include sensor data and image frame data;
[0066] The sensor data is acquired by the sensor acquisition module, and the sensor data includes soil moisture, soil temperature, soil pH and external illumination. The sensor acquisition module includes a humidity sensor, a temperature sensor, a pH sensor and a photosensitive sensor. The soil moisture is acquired by the humidity sensor, the soil temperature is acquired by the temperature sensor, the soil pH is acquired by the pH sensor, and the external illumination is acquired by the photosensitive sensor. The image frame data includes a single frame image and a multi-frame image, and the image frame data is acquired by a monitoring camera.
[0067] Among them, a single-frame image represents a frame of image, and a multi-frame image represents multiple frames of image. The multi-frame image includes a single-frame image, and several preceding frames of image and several succeeding frames of image corresponding to the single-frame image, which are collectively referred to as context image frames of the single-frame image.
[0068] At the corresponding time t2, different types of ground film related monitoring data are collected, and the number j corresponding to the current collection device area is bound to the ground film related monitoring data, so as to obtain the ground film related monitoring data of several collection device areas, and the number j is used as the partition identifier.
[0069] Specifically, the process of fusing different types of ground film related monitoring data to generate ground film comprehensive monitoring data includes:
[0070] Traverse the partition identifier j, use different types of ground film related monitoring data corresponding to each partition identifier as the data object of data fusion, and convert the data format of the data object into a preset data fusion standard format;
[0071] The sensor data in the monitoring data related to ground film converted into the data fusion standard format is used as the fusion element 1, and the image frame data in the monitoring data related to ground film converted into the data fusion standard format is used as the fusion element 2;
[0072] The fusion element one is marked as Rh1, and the fusion element two is marked as Rh2. Before data fusion is performed on the fusion element one Rh1 and the fusion element two Rh2, different types of feature parameters are extracted from Rh1 and Rh2 respectively;
[0073] The types of feature parameter extraction include feature curve segmentation labeling and feature image hierarchical labeling, wherein the feature curve segmentation labeling is performed on the fusion element 1 Rh1, and the feature image hierarchical labeling is performed on the fusion element 2 Rh2;
[0074] The content of the characteristic curve segmentation mark is as follows: establish a Cartesian coordinate system, take the values corresponding to the changes in soil moisture, soil temperature, soil pH and external illumination in the fusion element 1 over time as the ordinates, and record them as y1, y2, y3 and y4, take each time node corresponding to the values that change over time as the abscissa, and record them as x1, x2, x3 and x4, and then summarize and generate their own coordinate pairs, and record their own coordinate pairs as P1, P2, P3 and P4, then P1 = (x1, y1), P2 = (x2, y2), P3 = (x3, y3) and P4 = (x4, y4);
[0075] Mapping the coordinate pairs P1, P2, P3 and P4 corresponding to soil moisture, soil temperature, soil pH and external illumination to a Cartesian coordinate system, and connecting them in sequence through smooth curves to generate corresponding characteristic curves, wherein the characteristic curves include a humidity change curve, a temperature change curve, a pH change curve and an illumination change curve;
[0076] Preset the standard value change curves corresponding to soil moisture, soil temperature, soil pH and external illumination, and overlap and map the humidity change curve, temperature change curve, pH change curve and illumination change curve with their respective standard value change curves;
[0077] Then, the curve overlap points corresponding to each characteristic curve are generated by overlapping mapping comparison, and the curve overlap points are used as dividing points, and the curve part of the characteristic curve that is higher than the corresponding standard value change curve is marked as the upper curve, and the curve part of the characteristic curve that is lower than the corresponding standard value change curve is marked as the lower curve;
[0078] Taking the overlapping point of the curve as the starting point, set the reasonable extension number corresponding to the upper curve and the lower curve respectively, and record them as D 上 and D 下 , and then summarize the reasonable extension number to generate a reasonable numerical interval, and record the reasonable numerical interval as Ω, and Ω=[D 下 , D 上 ];
[0079] Mark all sensor data in the reasonable value interval Ω in the characteristic curve as cold monitoring data of fusion element one, and mark all sensor data not in the reasonable value interval Ω as hot monitoring data of fusion element one;
[0080] It should be noted that after each characteristic curve is overlapped and mapped with its respective standard value change curve, the number of generated curve overlap points is 2 by default. Connect the two curve overlap points to generate a reference line segment, and make a perpendicular line segment to the reference line segment. The intersection of the reference line segment and the perpendicular line segment is the origin of the coordinates. The origin is recorded as (0, 0). Then the reasonable extension number D 上 and D 下 The coordinates on the vertical line segment are (0, D 上 ) and (0, D 下 ), where the monitoring data related to ground film within the reasonable value interval Ω is normal data, and its priority for analysis is low, while the monitoring data related to ground film not within Ω is abnormal data, and its priority for analysis is high, so that abnormal problems can be analyzed more quickly;
[0081] The content of the hierarchical annotation of the feature image is as follows: the single-frame image and the multi-frame image in the fusion element 2 Rh2 are used as the main feature image and the auxiliary analysis feature image respectively, and the main feature image is hierarchically annotated to obtain the feature items that need to be annotated, and the feature items include the degree of damage to the mulch film, the crop pest situation and the crop growth situation;
[0082] The hierarchical annotation includes a first-layer annotation and a second-layer annotation. The first-layer annotation is performed on the main feature image to locate the approximate feature content of the preliminary feature item. The context image frame corresponding to the single frame of the main feature image is pulled out by auxiliary analysis of multiple frames corresponding to the feature image. Then, the specific feature details corresponding to the feature item are further obtained by performing a second-layer annotation on the context image frame. The obtained specific feature details are used as the image feature data of the second fusion element.
[0083] Set a data text, the data content of which is initially empty, enter the cold monitoring data and the hot monitoring data corresponding to the fusion element one into the data text, enter the image feature data corresponding to the fusion element two into the data text, and then generate different text fields corresponding to the fusion element one and the fusion element two, associate the same parts of the two text fields and use them as a text association item, and do not perform any operation on the different parts of the two text fields;
[0084] Data fusion is performed on several text-related items, and then comprehensive monitoring data of ground film is generated through data fusion. The content of data fusion is as follows: the modal dimension coefficient of each text-related item is extracted through a multimodal data fusion algorithm, which is denoted as G, and a modal dimension interval is set. The modal dimension interval includes a high-frequency modal interval and a general modal interval, which are denoted as GP and YB respectively. When G∈GP, all text-related items corresponding to G are summarized to generate the main characteristic mode of ground film. When G∈YB, all text-related items corresponding to G are summarized to generate the common characteristic mode of ground film. If neither G∈GP nor G∈YB is true, the modal dimension coefficient of the current text-related item is extracted again through a multimodal data fusion algorithm, and the main characteristic mode of ground film and the common characteristic mode of ground film are spliced and fused to generate comprehensive monitoring data of ground film.
[0085] It should be noted that the modal dimension coefficient is the proportion of each text-related item in the total number of text-related items, among which the high-frequency modal interval represents the main numerical interval reflecting the characteristics of ground film-related data, and the general modal interval represents other numerical intervals reflecting the characteristics of ground film-related data.
[0086] Specifically, the process of modular association and multi-path dynamic transmission of comprehensive monitoring data of ground film includes:
[0087] Set a number of module nodes, each module node corresponds to a data transmission protocol and a data error correction code, number the module nodes and record the number as k, k = 1, 2, 3, ..., q, where q is a natural number greater than 0;
[0088] The amount of data obtained for comprehensive monitoring of ground film is denoted as C 传输 , distribute the comprehensive monitoring data of ground film to several module nodes, and the amount of comprehensive monitoring data of ground film distributed to the module node numbered k is C[k] 分配 , the transmission capacity corresponding to the module node numbered k is recorded as C[k] 承载 , then the following relationship is satisfied: C[k] 分配 =C[k] 承载 ;
[0089] It should be noted that there is C[1] 分配 +C[2]分配 +C[3] 分配 +……+C[k] 分配 +……+C[q] 分配 =C 传输 , the transmission capacity corresponding to all module nodes is expressed as: C[1] 承载 +C[2] 承载 +C[3] 承载 +……+C[k] 承载 +C[q] 承载 ;
[0090] When each module node receives the allocated comprehensive monitoring data of ground film, the modular association of all module nodes is successful. Otherwise, the comprehensive monitoring data of ground film will continue to be allocated to the module nodes that have not received it.
[0091] Read the data transmission protocol corresponding to the module node numbered k, and then establish a transmission path between the current module node numbered k and the data analysis module, repeat the above operation, and then establish several transmission paths, each of which allows data transmission of the corresponding part of the ground film comprehensive monitoring data of the adjacent transmission path under the set dynamic transmission conditions;
[0092] The comprehensive monitoring data of the ground film transmitted through each transmission path is marked as segmented transmission data, and the dynamic transmission conditions are set as follows: if the real-time packet loss rate, transmission signal strength and delay rate of the current transmission path are all lower than the respective preset qualified thresholds, the segmented transmission data of the current transmission path will be transferred to the adjacent transmission path, and the data will be transmitted by the adjacent transmission path;
[0093] The data error correction code is used to correct the segmented transmission data corresponding to each transmission path, and determine whether the data error correction code has changed. If so, the data transmission of the segmented transmission data corresponding to the data error correction code is re-performed, if not, no operation is performed.
[0094] Specifically, the process of receiving the comprehensive monitoring data of ground film from multi-path dynamic transmission and integrating the comprehensive monitoring data of ground film from different transmission paths includes:
[0095] Set up a data integration program, and set up a program self-check period, a data receiving period, and a data integration period of the path integration program accordingly; perform a program self-check of the data integration program during the program self-check period; receive ground film comprehensive monitoring data from multi-path dynamic transmission during the data receiving period; and integrate segmented transmission data corresponding to different transmission paths during the data integration period;
[0096] The program self-check content is as follows: set a sample data set, the sample data set includes different grouped sample data, each grouped sample data is associated with a corresponding virtual path, the sample data set is input into the data integration program, the data integration program sets any virtual path as the integration path, integrates the grouped sample data according to the integration path, and determines whether the integration is successful. If there is only one virtual path after the integration, the integration is successful, otherwise, the integration fails;
[0097] An encrypted receiving space is set up during the data receiving period, the segmented transmission data corresponding to different transmission paths are received into the encrypted receiving space, and the data in the encrypted receiving space is encrypted, thereby generating a corresponding space key;
[0098] The content of integrating the segmented transmission data is as follows: obtaining the decryption key corresponding to the space key, and then decrypting the segmented transmission data in the encrypted receiving space by using the decryption key, generating a corresponding data integration node according to each transmission path, and storing the segmented transmission data corresponding to the transmission path in the data integration node;
[0099] Each data integration node is provided with a corresponding integration point, and the corresponding integration points of every two data integration nodes are connected in turn, and then all the data integration nodes are connected. When all the data integration nodes are successfully connected, all the segmented transmission data are synchronously integrated into the comprehensive monitoring data of the ground film.
[0100] Specifically, the process of analyzing the integrated ground film comprehensive monitoring data and generating data analysis reports includes:
[0101] Acquire the integrated comprehensive monitoring data of ground film, and import it into a set data analysis program for data analysis, wherein the data analysis program is pre-loaded with different types of report field sets, and each type of report field set is associated with a section of ground film analysis data content;
[0102] A bag-of-words model is set up, wherein the bag-of-words model is used to perform word class grouping and word frequency statistics on the comprehensive monitoring data of ground film, and the integrated comprehensive monitoring data of ground film is input into the bag-of-words model as model data;
[0103] The bag-of-words model performs word class grouping according to the model data, and then generates several different monitoring data vocabulary groups by grouping. The detailed content of the word class grouping is as follows: the bag-of-words model is provided with several different word class clusters, each word class cluster is used to group one type of monitoring data vocabulary group, and all the vocabulary contents corresponding to the model data are obtained, and then the vocabulary contents belonging to the same word class cluster are grouped into one type of monitoring data vocabulary group, and the operation is repeated to group several types of monitoring data vocabulary groups;
[0104] The monitoring data vocabulary group includes several monitoring data vocabulary. The frequency of each monitoring data vocabulary is obtained by word frequency statistics, and the word frequency is recorded as f. The abnormal word frequency interval is set and recorded as Ω 词频 , if f∈Ω 词频 , then mark the word frequency as abnormal word frequency, and remove the monitoring data vocabulary corresponding to the abnormal word frequency. If f Ω 词频 , it means that the word frequency is normal and no operation is performed;
[0105] The monitoring data vocabulary included in the monitoring data vocabulary group after the elimination operation is completed is marked as a monitoring field, the report field set includes a plurality of report fields, and the monitoring data vocabulary group and the report field set are matched to obtain a matching rate, and the matching rate is recorded as R;
[0106] Then R=(Z 匹 / Z 总 )×100%, where Z 匹 Indicates the number of successful matches between the monitoring fields in the monitoring data vocabulary group and the report fields in the report field set, Z 总 Indicates the total number of report fields in the report field set, sets the matching achievement rate, recorded as R`, and sets blank report text, which is used as a data recording carrier for the data analysis report;
[0107] When R≥R`, a section of mulch analysis data content associated with the current report field set is used as the data to be filled in the report. When R<R`, no operation is performed, and all the report data to be filled that meet the condition R≥R` are recorded in the blank report text, and then the blank report text is converted into a data analysis report.
[0108] Specifically, the process of comparing the data analysis report with the preset alarm threshold and then determining whether to generate an alarm prompt information according to the comparison result includes:
[0109] The data analysis report includes a plurality of report items, each of which is provided with a corresponding project indicator, and the plurality of report items are numbered, and the number is recorded as e, where e=1, 2, 3, ..., v, wherein v is a natural number greater than 0;
[0110] Obtain the project indicator corresponding to the report item numbered e from the data analysis report, and then obtain the indicator risk measurement coefficient corresponding to the project indicator, and record the indicator risk measurement coefficient as γ[e], set a corresponding alarm threshold for the project indicator corresponding to the report item numbered e, and record the alarm threshold as τ[e];
[0111] Compare the numerical values of γ[e] and τ[e], and generate different comparison results, and then decide whether to generate an alarm prompt information according to the comparison results, wherein the comparison results include result one and result two;
[0112] When γ[e]≥τ[e], the comparison result generated is result 1, and the content of result 1 is: the numerical value of the project indicator corresponding to the current report item is preliminarily analyzed to be abnormal;
[0113] When γ[e]<τ[e], the comparison result generated is result 2, and the content of result 2 is: the numerical value of the project indicator corresponding to the current report item is preliminarily analyzed to be within the normal range;
[0114] When the comparison result is result one, an alarm prompt message is generated. The content of the alarm prompt message is: there is an abnormal risk in the project indicator corresponding to the current report item. Please arrange relevant risk elimination personnel to conduct risk management of the corresponding part of the mulch area as soon as possible; when the comparison result is result two, no alarm prompt message is generated.
[0115] Specifically, the process of determining whether to perform feedback change of the alarm threshold according to the alarm success rate of the alarm prompt information includes:
[0116] Get the alarm success rate of the alarm prompt information, record the alarm success rate as Sc, then Sc=(N 成功 / N 总 )×100%, where N 成功 N is the number of times the alarm prompt information successfully completes the alarm prompt. 总 The total number of times the alarm prompt information is generated;
[0117] Set the feedback change interval of the alarm threshold. According to the subordinate relationship between the alarm success rate Sc and the feedback change interval, decide whether to change the alarm threshold feedback. The feedback change interval is Ω. 反馈 , there is Ω 反馈 =[0%,u], where u is set to 70%, which means the minimum limit value of the alarm success rate. The value of u can be changed;
[0118] When Sc∈Ω 反馈 When , the alarm threshold is changed;
[0119] When Sc Ω 反馈 , no feedback change of the alarm threshold is performed;
[0120] The content of the feedback change is as follows: Feedback changes include regular changes, event-driven changes and custom changes, among which the regular changes are as follows: the alarm threshold set for each item indicator in the report item is numerically increased or decreased until the alarm success rate is not within the feedback change range, so as to improve the alarm success rate;
[0121] The content of event-driven changes is as follows: set change events, which include external attacks, firewall information errors, and system crashes. When a corresponding change event occurs, it means that the accuracy of the current alarm prompt information is likely to decrease. The alarm threshold of each project indicator is reduced, and u is increased to facilitate faster alarm warning;
[0122] The content of the customized changes is as follows: the user increases or decreases the alarm threshold of each project indicator in the report item according to personal monitoring intensity requirements of the monitored mulch area. When the alarm threshold of each project indicator is reduced, it means that the user's monitoring intensity requirement is strengthened. Conversely, when the alarm threshold of each project indicator is increased, it means that the user's monitoring intensity requirement is weakened.
[0123] Specifically, the process of reading the alarm prompt information and the equipment parameter information stored in the monitoring center, and then executing the corresponding ground film automatic management measures according to the alarm prompt information and the equipment parameter information includes:
[0124] The data reading unit reads the alarm prompt information, establishes data communication between the data reading unit and the monitoring center, and then obtains the device parameter information stored in the monitoring center, and transmits the alarm prompt information and the device parameter information to the automation management unit;
[0125] The equipment parameter information includes feeding automation parameter information and exhaust automation parameter information;
[0126] Execute different ground film automation management measures according to the alarm prompt information and the equipment parameter information, wherein the ground film automation management measures include fault management measures and ground film operation management measures, wherein the fault management measures are generated by the alarm prompt information, the ground film operation management measures are generated by the equipment parameter information, and the ground film operation management measures include feeding measures and exhaust measures;
[0127] A solution library is set up, in which specific implementation plans of several mulch film operation management measures are stored, and the feeding automation parameter information and the exhaust automation parameter information are input into the solution library, and then the specific implementation plans of the corresponding feeding measures and exhaust measures in the solution library are matched and obtained;
[0128] The automatic management unit starts the automatic feeding equipment and automatic exhaust equipment in the monitored mulch film area according to the specific implementation plan of the feeding measures and exhaust measures, and then implements the corresponding automatic feeding and automatic exhaust operations;
[0129] The fault management measures include fault handling priorities, a list of fault-related handling tools, and a list of fault handling personnel. The relevant personnel included in the fault handling personnel list are arranged to carry the fault handling tools corresponding to the fault-related handling tool list, and perform fault handling in the monitored mulch area in order of fault handling priority from high to low.
[0130] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.
Claims
1. The automatic monitoring system of ground film based on multimodal data fusion includes a monitoring center, It is characterized in that The monitoring center is communicatively connected with a data acquisition module, a data processing and fusion module, a data analysis module, an alarm prompt module and a ground film automation management module; The data acquisition module includes an equipment deployment unit and a multi-source data acquisition unit; the equipment deployment unit is used to deploy ground film related working equipment and ground film related acquisition equipment; the multi-source data acquisition unit is used to start the data acquisition work of the ground film related acquisition equipment, and then obtain different types of ground film related monitoring data; The data processing and fusion module is used to fuse different types of ground film related monitoring data, thereby generating ground film comprehensive monitoring data, and performing modular association and multi-path dynamic transmission for the ground film comprehensive monitoring data; The data analysis module is used to receive the comprehensive monitoring data of ground film from multi-path dynamic transmission, integrate the comprehensive monitoring data of ground film from different transmission paths, perform data analysis on the integrated comprehensive monitoring data of ground film, and then generate a data analysis report; The alarm prompt module compares the data analysis report with the preset alarm threshold, and then determines whether to generate an alarm prompt message according to the comparison result, and determines whether to perform feedback change of the alarm threshold according to the alarm success rate of the alarm prompt message; The mulch film automation management module is provided with a data reading unit and an automation management unit; the data reading unit is used to read the alarm prompt information and the equipment parameter information stored in the monitoring center; The automation management unit is used to execute corresponding ground film automation management measures according to the alarm prompt information and equipment parameter information; The process of generating comprehensive monitoring data for mulch film includes: Different types of ground film related monitoring data are used as data objects for data fusion, and the data format of the data objects is converted into a preset data fusion standard format. The sensor data and image frame data after the conversion are used as fusion element one and fusion element two, respectively, and are marked as Rh1 and Rh2, respectively. Before data fusion of Rh1 and Rh2, different types of feature parameters are extracted for Rh1 and Rh2, respectively. The types of feature parameter extraction include feature curve segmentation marking and feature image hierarchical annotation. Feature curve segmentation marking is performed on Rh1, and feature image hierarchical annotation is performed on Rh2. Generate characteristic curves and reasonable value intervals through segmented marking of characteristic curves, mark sensor data in the reasonable value interval of the characteristic curve as cold monitoring data of fusion element one, mark sensor data not in the reasonable value interval as hot monitoring data of fusion element one, and obtain image feature data of fusion element two through hierarchical annotation of characteristic images; Set the data text, enter the cold monitoring data and the hot monitoring data corresponding to the fusion element one into the data text, enter the image feature data corresponding to the fusion element two into the data text, and then generate different text fields corresponding to the fusion element one and the fusion element two, associate the same parts of the text fields in the fusion element one and the fusion element two, and use them as a text association item, do not perform any operation on the different parts of the text fields in the fusion element one and the fusion element two, perform data fusion on several text association items, and then generate comprehensive monitoring data of ground film through data fusion; The process of modular association and multi-path dynamic transmission of comprehensive monitoring data of ground film includes: Set up several module nodes, each module node corresponds to a data transmission protocol, and number the several module nodes and record them as k, k = 1, 2, 3, ..., q, where q is a natural number greater than 0. The amount of data of the comprehensive monitoring data of the ground film is obtained, which is recorded as C 传输 , distribute the comprehensive monitoring data of ground film to several module nodes, and the amount of comprehensive monitoring data of ground film distributed to the module node numbered k is C[k] 分配 , the transmission capacity corresponding to the module node numbered k is recorded as C[k] 承载 , we have C[k] 分配 =C[k] 承载 When each module node receives the allocated comprehensive monitoring data of ground film, the modular association of all module nodes is successful. Otherwise, the comprehensive monitoring data of ground film will continue to be allocated to the module nodes that have not received it. Read the data transmission protocol corresponding to the module node numbered k, and then establish a transmission path between the current module node numbered k and the data analysis module, repeat the above operation, and then establish several transmission paths, each of the transmission paths is allowed to replace the data transmission of the corresponding part of the ground film comprehensive monitoring data of the adjacent transmission path under the set dynamic transmission conditions, and the ground film comprehensive monitoring data transmitted through each transmission path is marked as segmented transmission data, and the dynamic transmission conditions are set as follows: if the real-time packet loss rate, transmission signal strength and delay rate of the current transmission path are all lower than the respective preset qualified thresholds, the segmented transmission data of the current transmission path is transferred to the adjacent transmission path, and the adjacent transmission path is used for data transmission; The process of integrating comprehensive monitoring data of ground film from different transmission paths includes: A data integration program is set, and a data receiving period and a data integration period corresponding to the data integration program are set, the ground film comprehensive monitoring data from multi-path dynamic transmission is received during the data receiving period, and the segmented transmission data corresponding to different transmission paths are integrated during the data integration period; During the data receiving period, an encrypted receiving space is set, and the segmented transmission data corresponding to different transmission paths are received into the encrypted receiving space, and the encrypted receiving space is encrypted to generate a corresponding space key, and a decryption key set corresponding to the space key is obtained, and then the segmented transmission data in the encrypted receiving space is decrypted and obtained by the decryption key, and a corresponding data integration node is generated according to each transmission path, and the segmented transmission data corresponding to the transmission path is stored in the data integration node. Each data integration node is provided with a corresponding integration point, and the corresponding integration points of every two data integration nodes are connected, and then all the data integration nodes are connected. When all the data integration nodes are successfully connected, all the segmented transmission data are synchronously integrated into the comprehensive monitoring data of the ground film; The process of analyzing the integrated ground film comprehensive monitoring data and generating data analysis reports includes: Obtain the integrated comprehensive monitoring data of ground film, and import it into the set data analysis program for data analysis. The data analysis program is pre-entered with different types of report field sets, and each type of report field set is associated with a section of ground film analysis data content; The bag-of-words model is set up to perform word-class grouping and word frequency statistics on the comprehensive monitoring data of ground film. The integrated comprehensive monitoring data of ground film is input into the bag-of-words model as model data. The bag-of-words model performs word-class grouping according to the model data, and then generates several different monitoring data vocabulary groups by grouping. The monitoring data vocabulary group includes several monitoring data vocabulary. The word frequency of each monitoring data vocabulary is obtained through word frequency statistics, which is recorded as f. The abnormal word frequency interval is set and recorded as Ω. 词频 , if f∈Ω 词频 , then mark the word frequency as abnormal word frequency, and remove the monitoring data vocabulary corresponding to the abnormal word frequency. If f Ω 词频 , it means that the word frequency is normal and no operation is performed; The monitoring data vocabulary included in the monitoring data vocabulary group after the elimination operation is marked as a monitoring field. The report field set includes several report fields. The monitoring data vocabulary group and the report field set are matched to obtain a matching rate. The matching rate is recorded as R, R = (Z 匹 / Z 总 )×100%, where Z 匹 Indicates the number of successful matches between the monitoring fields in the monitoring data vocabulary group and the report fields in the report field set, Z 总 Indicates the total number of report fields in the report field set, sets the matching achievement rate, denoted as R`; When R≥R`, a section of mulch analysis data corresponding to the current report field set is used as the report data to be filled in. When R<R`, no operation is performed, and all the report data to be filled in that meet the condition R≥R` are recorded in the set blank report text, and then the blank report text is converted into a data analysis report; The process of comparing the data analysis report with the preset alarm threshold and then deciding whether to generate an alarm prompt information according to the comparison result includes: The data analysis report includes a number of report items, each of which is provided with a corresponding project indicator. The number of report items is numbered and recorded as e, where e=1, 2, 3, ..., v, where v is a natural number greater than 0. The project indicator corresponding to the report item numbered e is obtained from the data analysis report, and then the indicator risk measurement coefficient corresponding to the project indicator is obtained and recorded as γ[e]. A corresponding alarm threshold is set for the project indicator corresponding to the report item numbered e and recorded as τ[e]; Compare the numerical values of γ[e] and τ[e], and generate different comparison results, and then decide whether to generate an alarm prompt information based on the comparison results. The comparison results include result one and result two. When γ[e]≥τ[e], the generated comparison result is result one, and the content of result one is: the numerical value of the project indicator corresponding to the current report item is initially analyzed to be abnormal; when γ[e]<τ[e], the generated comparison result is result two, and the content of result two is: the numerical value of the project indicator corresponding to the current report item is initially analyzed to be within the normal range; when the comparison result is result one, an alarm prompt information is generated, and when the comparison result is result two, no alarm prompt information is generated.
2. According to claim 1, the automatic monitoring system for ground film based on multimodal data fusion, It is characterized in that The process of deploying ground film related working equipment and ground film related collection equipment includes: Select the mulch area that needs to be monitored, select and schedule mulch-related working equipment and mulch-related collection equipment, and obtain the working coverage area of mulch-related working equipment, recorded as S 工设 , obtain the collection coverage area of the ground film related collection equipment, recorded as S 采集 , the area of the ground film area to be monitored is recorded as S, according to S and S 工设 A number of working equipment areas are divided on the ground film area and numbered as i, where i = 1, 2, 3, ..., n. 采集 A number of collection equipment areas are divided and numbered as j, where j = 1, 2, 3, ..., m, where n and m are natural numbers greater than 0; In the working equipment areas numbered i and the collecting equipment areas numbered j, ground film-related working equipment and ground film-related collecting equipment are deployed respectively, and their corresponding communication sequences are obtained, which are recorded as A[i] and B[i] respectively. The ground film-related working equipment includes automatic feeding equipment and automatic exhaust equipment, and the ground film-related collecting equipment includes sensor collection modules and monitoring cameras. The equipment parameter information of the ground film-related working equipment is obtained and transmitted to the monitoring center for storage.
3. According to claim 2, the automatic monitoring system for ground film based on multimodal data fusion, It is characterized in that The process of obtaining monitoring data related to different types of mulch films includes: Set the communication link time and data collection period, and record the data collection period as T 采集 , there is T 采集 =[t1, t2], where t1 is the start time of the data collection period, and t2 is the end time of the data collection period. At the communication link time, the communication sequences A[i] and B[i] are obtained, and then the data communication with the film-related working equipment corresponding to A[i] and the film-related collection equipment corresponding to B[i] is established. At the time point corresponding to t1, the data collection work of the film-related collection equipment is started, and then different types of film-related monitoring data are obtained. The types of film-related monitoring data include sensor data and image frame data; Among them, the sensor data is collected and acquired by the sensor acquisition module. The sensor data includes soil moisture, soil temperature, soil pH and external illumination. The image frame data includes single-frame images and multi-frame images. The image frame data is collected by the monitoring camera. At the corresponding time t2, the monitoring data related to different types of ground films are collected. The number j corresponding to the current collection device area is bound to the ground film related monitoring data, and then the ground film related monitoring data of several collection device areas are obtained.
4. According to claim 3, the automatic monitoring system for ground film based on multimodal data fusion, It is characterized in that The process of determining whether to change the alarm threshold feedback according to the alarm success rate of the alarm prompt information includes: Get the alarm success rate of the alarm prompt information, recorded as Sc, set the feedback change interval of the alarm threshold, and record it as Ω 反馈 , there is Ω 反馈 =[0%,u], where u is set to 70% and can be changed; When Sc∈Ω 反馈 When , the alarm threshold is changed; When Sc Ω 反馈 , no feedback change of the alarm threshold is performed; The feedback changes include regular changes, event-driven changes, and custom changes.
5. According to claim 4, the automatic monitoring system for ground film based on multimodal data fusion, It is characterized in that The process of reading the alarm prompt information and the equipment parameter information stored in the monitoring center and then executing the corresponding mulch film automation management measures includes: The data reading unit reads the alarm prompt information, establishes data communication between the data reading unit and the monitoring center, and then obtains the device parameter information stored in the monitoring center, and transmits the alarm prompt information and the device parameter information to the automation management unit; The equipment parameter information includes feeding automation parameter information and exhaust automation parameter information; Execute different ground film automation management measures according to the alarm prompt information and equipment parameter information. The ground film automation management measures include fault management measures and ground film operation management measures. Among them, the fault management measures are generated by the alarm prompt information, and the ground film operation management measures are generated by the equipment parameter information. The ground film operation management measures include feeding measures and exhaust measures; A solution library is set up, in which specific implementation plans of several mulch film operation management measures are stored, and the feeding automation parameter information and the exhaust automation parameter information are input into the solution library, and then the specific implementation plans of the corresponding feeding measures and exhaust measures in the solution library are matched and obtained; The automatic management unit starts the automatic feeding equipment and automatic exhaust equipment in the monitored mulch film area according to the specific implementation plan of the feeding measures and exhaust measures, and then implements the corresponding automatic feeding and automatic exhaust operations; The fault management measures include fault handling priorities, a list of fault-related handling tools, and a list of fault handling personnel. The relevant personnel included in the fault handling personnel list are arranged to carry the fault handling tools corresponding to the fault-related handling tool list, and perform fault handling in the monitored mulch area in order of fault handling priority from high to low.
