Industrial production equipment abnormity tracing method and system based on cloud side-end cooperation

Abstract

The invention discloses an industrial production equipment abnormity traceability method and system based on cloud side-end cooperation, and relates to the technical field of industrial equipment abnormity traceability. The edge side obtains the operation data of the industrial production equipment in real time and performs anomaly detection, an anomaly source is locked through the abnormal operation data, and if the edge side can lock the anomaly source, the operation data of the industrial production equipment is not transmitted to the cloud; the cloud analyzes the abnormal source according to the priority according to the operation data transmitted by the edge side, and feeds back abnormal source information to the corresponding edge side after determining the abnormal source; the delay of operation data processing is reduced through the edge side, so that the abnormal condition can be quickly detected by the edge side; abnormal conditions are traced through cooperation of the cloud and the edge side, the abnormal condition traceability speed is increased, and adverse effects caused by the abnormal conditions are reduced. The data flow is effectively managed, and the use of network resources is optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of industrial equipment anomaly tracing, and specifically to an industrial production equipment anomaly tracing method and system based on cloud-edge-end collaboration. Background Art

[0002] Cloud-Edge-End is a distributed computing architecture that disperses data processing and analysis tasks among cloud computing centers, edge computing nodes, and terminal devices to optimize the response speed, reliability, and data processing capabilities of the entire system. In industrial production, tracing the source of equipment anomalies is a key issue that directly affects production efficiency and product quality. Combining the powerful data processing capabilities of cloud computing and the low latency characteristics of edge computing, efficient anomaly detection and accurate fault location can be achieved. After edge computing detects an abnormality, the powerful data processing capabilities of cloud computing can be used to trace the source of the anomaly. However, the edge computing nodes themselves also have data processing capabilities. When the edge computing itself can determine the source of the anomaly, continuing to transmit data to the cloud will result in a waste of resources. For this reason, how to improve the resource utilization of the cloud-edge end in the process of tracing the source of equipment anomalies has become an urgent problem that needs to be solved. Summary of the invention

[0003] The purpose of the present invention is to provide an industrial production equipment anomaly tracing method and system based on cloud-edge-end collaboration to solve the problems raised in the prior art.

[0004] To achieve the above purpose, the present invention provides the following technical solution: an abnormality tracing method for industrial production equipment based on cloud-edge-end collaboration, comprising: S11, acquiring historical operation data of industrial production equipment, and establishing an equipment database based on the historical operation data of the industrial production equipment; S12, the edge side acquires the operation data of the industrial production equipment in real time and performs anomaly detection, and locks the abnormal source through the abnormal operation data. If the edge side cannot lock the abnormal source, the operation data of the industrial production equipment is transmitted to the cloud, and the process goes to step S13; if the edge side can lock the abnormal source, the operation data of the industrial production equipment is not transmitted to the cloud, and the edge side processes the abnormal source, and the process ends; S13, the cloud analyzes the abnormal source according to the priority based on the operation data transmitted by the edge side, and after determining the abnormal source, it feeds back the abnormal source information to the corresponding edge side, which processes the abnormal source and ends. In step S11, the establishment of the equipment database based on the historical operation data of the industrial production equipment further includes the following steps: S21, obtaining historical operation data of the industrial production equipment when it is working normally and generating a standard trend graph of each historical operation data; S22, obtaining historical operating data of each abnormal source that deviates from the standard trend graph when an abnormal situation occurs, establishing a connection between the historical operating data that deviates from the standard trend graph and the abnormal source, and generating an abnormal trend graph of each abnormal source based on the operating data that deviates from the standard trend graph, wherein the abnormal trend graph of each abnormal source is a trend graph of part of the operating data or a trend graph of all the operating data; S23, when an abnormal source occurs, the operation data that deviates from the standard trend graph is defined as the relevant operation data, and the edge side responsible for the relevant operation data is defined as the relevant edge side; for the i-th relevant edge side, the response time t of other relevant edge sides is obtained, t=t i +t j , where t i is the time required for the i-th related edge side to obtain information from the cloud, t j The time required for other edge sides to send information to the cloud.

[0005] When an abnormal source occurs, it will affect the operating data of industrial production equipment. The affected operating data of industrial production equipment may be monitored by multiple different edge sides. Therefore, one abnormal source corresponds to multiple operating data, and these multiple operating data are monitored by multiple different edge sides. These responsible edge sides are the relevant edge sides of the abnormal source. By obtaining the trend chart of the operating data during normal operation and comparing the current operating data curve with the trend chart, it can be determined whether the operating data is normal; similarly, by obtaining the abnormal trend chart of the operating data when the abnormal source occurs and comparing the current operating data curve with the abnormal trend chart, it can be determined whether the current operating data matches the abnormal source.

[0006] In step S12, the edge side acquires the operation data of the industrial production equipment in real time and performs anomaly detection, and further includes the following steps: S30, let Ak represent the kth edge side, the edge side Ak compares the acquired operation data curve with the standard trend graph of the corresponding operation data, calculates the root mean square error between the operation data and the corresponding standard trend graph, and if the root mean square error is greater than a threshold, determines that the operation data is abnormal operation data; The method of locking the abnormal source through abnormal operation data comprises the following steps: S31, the abnormal operation data set determined by the edge side Ak is Mk, the number of operation data contained in the abnormal operation data set Mk is mk, and from all abnormal sources, the abnormal sources that are simultaneously connected to all the operation data in the abnormal operation data set Mk are screened out, and the screened abnormal sources are set as target abnormal sources. If the number of target abnormal sources is 1, the abnormal sources are locked, and the edge side Ak processes the abnormal sources, and the process ends; if the number of target abnormal sources is not 1, the process proceeds to step S32; S32, obtaining the operation data set Na associated with the a-th target abnormal source, calculating the consistency css(k, a) of the abnormal operation data between the a-th target abnormal source and the edge side Ak, css(k, a) = (Mk∩Na) / (Mk∪Na), Mk∩Na represents the number of identical operation data in the set Mk and the set Na, and Mk∪Na represents the total number of non-repeated operation data in the set Mk and the set Na; When an abnormal source appears, it affects multiple operating data. The affected operating data deviates from the normal trend chart and is detected by multiple edge sides. For a single edge side, it may not be responsible for monitoring all the affected operating data; when an abnormal source appears, some of the operating data monitored by the edge side Ak is judged to be abnormal operating data. At this time, the edge side Ak begins to trace the abnormality; according to the operating data judged as abnormal row data, combined with the abnormal sources that will simultaneously affect the operating data judged as abnormal row data, the target abnormal source can be obtained. When the number of target abnormal sources is 1, the abnormal source can be determined without transmitting data to the cloud, saving resources; when the target abnormal source is not 1, according to the operating data judged as abnormal row data, the consistency of the abnormal operating data of the edge side Ak and each target abnormal source can be determined; For example, for the ath target abnormal source, a total of na abnormal operation data are included, and the edge side Ak detects mk abnormal operation data. When the na abnormal operation data are the same as the mk abnormal operation data, the consistency of the abnormal operation data of the edge side Ak and the ath target abnormal source is 1, because when the target abnormal source appears, all abnormal operation data are detected by the edge side Ak, and the abnormal operation data detected by the edge side Ak are all related to the ath target abnormal source; wherein the na abnormal operation data and the mk abnormal operation data are the same means that the operation parameters involved are the same, such as mk abnormal operation data are the voltages at c1, c2, and c3, and the na abnormal operation data are also the voltages at c1, c2, and c3, then the na abnormal operation data are the same as the mk abnormal operation data, and the actual voltage value is not considered; when the na abnormal operation data and the mk abnormal operation data are not exactly the same, the consistency is determined according to the proportion of the same abnormal operation data in all abnormal operation data; Here, only abnormal operation data is analyzed, and whether the abnormal operation data matches is not considered. Even if they are all abnormal operation data, the trends shown may be different. Therefore, it is necessary to determine the matching degree. In order to find the abnormal source as soon as possible, all possible abnormal sources are eliminated in the order of matching degree. At the same time, the feedback information from the cloud is monitored to determine whether the operation data needs to be transmitted to the cloud. S33, determine the matching degree between the a-th target abnormal source and the abnormal operation data of the edge side Ak in order of consistency from large to small, and check the abnormal source according to the matching degree from large to small; let Db represent the b-th operation data in the abnormal operation data set Mk, calculate the root mean square error between the curve of the operation data Db obtained by the edge side Ak and the abnormal trend chart of the operation data Db of the a-th target abnormal source, if the root mean square error is less than the threshold, then it is judged that the operation data Db of the edge side Ak matches the operation data Db of the a-th target abnormal source, determine the operation data set Rk that matches the a-th target abnormal source in the abnormal operation data set Mk of the edge side Ak, let the number of operation data contained in Rk be num(k, a), and then obtain the matching degree fit(k, a) between the a-th target abnormal source and the abnormal operation data of the edge side Ak, fit(k, a)=num(k, a) / (Mk∪Na).

[0007] Compared with consistency, a judgment is added on whether the operation data curve matches the abnormal trend graph of the operation data Db of the ath target abnormal source, and the value of the matching degree will be less than or equal to the consistency.

[0008] In step S30, the step of calculating the root mean square error between the operating data and the corresponding standard trend graph further includes the following steps: S41, aligning the sampling nodes of the standard trend graph with the sampling nodes of the operating data curve acquired by the edge side Ak, and calculating the root mean square error between the operating data curve and the corresponding standard trend graph at the sampling nodes; S42, translate the standard trend graph by a fixed time step step, recalculate the root mean square error between the operating data curve and the corresponding standard trend graph at the sampling node of the edge side Ak, the value of the sampling time point of the standard trend graph at the edge side Ak after translation is px, and the root mean square error is calculated according to px and the value of the operating data curve at the sampling node of the edge side Ak; px=x+sign×step×v, where v represents the rate of change of the operating data on the standard trend graph, the sampling node of the edge side Ak is between the two sampling nodes of the standard trend graph after translation, and the rate of change of the operating data is obtained according to the two sampling nodes of the standard trend graph and the sampling period of the edge side Ak; when the standard trend graph is translated backward on the time axis, step is 1, and when the standard trend graph is translated forward on the time axis, sign is -1; x is the value of the operating data at the sampling node of the edge side Ak with the standard trend graph before translation; S43, each time translating a time step, when the root mean square error reaches a minimum value, determining the root mean square error between the operating data and the corresponding standard trend graph; In order to determine the root mean square error, it is necessary to align the operating data curve with the corresponding standard trend chart, that is, to translate the standard trend chart on the time axis. For the same variable, two identical data acquisition devices are used for data acquisition, but the two data acquisition devices start working at different times. The same variable curves obtained by the two data acquisition devices are approximately the same, but directly aligning the curves obtained by the two data acquisition devices according to the sampling nodes cannot achieve the purpose of aligning the two curves. Therefore, it is essential to translate the standard trend chart on the time axis.

[0009] In step S33, the calculation of the root mean square error between the curve of the operating data Db obtained by the edge side Ak and the abnormal trend graph of the operating data Db of the a-th target abnormal source also includes the following steps: translating the abnormal trend graph of the operating data Db of the a-th target abnormal source in accordance with steps S41, S42 and S43 and calculating the root mean square error after each translation; when the root mean square error reaches the minimum value, determining the root mean square error between the curve of the operating data Db obtained by the edge side Ak and the abnormal trend graph of the operating data Db of the a-th target abnormal source.

[0010] In step S12, the following steps are also included: When the edge side Ak determines that there is abnormal operation data through judgment, if the edge side Ak can lock the abnormal source, the edge side Ak sends the information of the determined abnormal source to the cloud. After the cloud receives the information of the determined abnormal source, it feeds back the information of the determined abnormal source to all edge sides; if the edge side Ak does not lock the abnormal source, and within the time threshold, does not receive the information of the determined abnormal source sent by the cloud, then the edge side Ak transmits the operation data of the industrial production equipment to the cloud, and the cloud determines the abnormal source, and at the same time transmits the consistency and matching degree of the abnormal operation data of the edge side Ak with all target abnormal sources to the cloud; the time threshold is determined by the following steps: the edge side Ak obtains the response time of the relevant edge side of the ath target abnormal source, takes the maximum value of all response times of the ath target abnormal source to obtain the response time of the ath target abnormal source; takes the maximum value of the response time of all target abnormal sources to obtain the time threshold.

[0011] After edge side Ak determines that there is abnormal operation data, if edge side Ak can lock the abnormal source, it needs to consider whether other edge sides can lock the abnormal source. If the abnormal source cannot be locked, other edge sides will transmit operation data to the cloud, and the cloud will trace the abnormal source. In fact, edge side Ak has locked the abnormal source, and there is no need to transmit operation data, consistency and matching analysis results to the cloud. Therefore, edge side Ak sends information about the determined abnormal source to the cloud, and the cloud feeds back information about the determined abnormal source to other edge sides. If other edge sides receive feedback information within the time threshold, there is no need to transmit operation data to the cloud, saving bandwidth resources. If edge side Ak cannot lock the abnormal source, it first analyzes the consistency and matching of the abnormal source, and tries to check possible abnormal sources. Before the time threshold is reached, the operation data is not transmitted. When the time threshold is reached, it means that other edge sides have also failed to lock the abnormal source and need to use the computing resources of the cloud platform. At this time, edge side Ak transmits the operation data, consistency and matching analysis results to the cloud.

[0012] In step S13, analyzing the abnormal source according to the priority also includes the following steps: The cloud determines the priority of the abnormal source based on the consistency and matching degree of the abnormal operation data received from each edge side and all target abnormal sources. For the ath target abnormal source, the priority ;W1 and W2 are weights;After determining the priorities of all target anomaly sources, the cloud analyzes the anomaly sources based on the operation data transmitted by the edge side in descending order of priority. For the ath target anomaly source, Represents the number of all abnormal operation data detected in the set Na, Represents the sum of the number of abnormal operation data detected by the set Na and all edge sides.

[0013] To achieve the above-mentioned objectives, the present invention provides the following technical solutions: an industrial production equipment abnormality tracing system based on cloud-edge-end collaboration, comprising a cloud side, an edge side and a terminal side, wherein the cloud side and the edge side are interconnected and are responsible for global operation data analysis. When the edge side cannot lock the abnormality source, the abnormality source is determined through large-scale data analysis to provide background support for abnormality tracing; the edge side and the terminal side are interconnected and are used to perform real-time analysis and abnormality judgment on the operation data of industrial production equipment, and lock the abnormality source; the terminal side is used to collect the operation data of industrial production equipment.

[0014] The terminal side also includes a data acquisition unit and a first communication unit, the data acquisition unit is used to obtain the operation data of the industrial production equipment, and the first communication unit is used to send the obtained operation data to the edge side. The edge side also includes a second communication unit, an abnormality detection unit, a consistency determination unit, a matching degree calculation unit, a synchronization unit, a second storage unit and an abnormal source processing unit; the second communication unit is used to realize data interaction between the edge side and the cloud; the abnormality detection unit is used to perform abnormality detection on the operation data responsible for the edge side; the consistency determination unit is used to determine the consistency between the abnormal operation data detected by the edge side and the abnormal source: the matching degree calculation unit is used to determine the matching degree between the abnormal operation data detected by the edge side and the abnormal source; the synchronization unit is used to align the standard trend chart and the abnormal trend chart with the operation data collected by the edge side; the second storage unit is used to store the standard trend chart, the abnormal trend chart of the operation data and the response time data of the edge side; the abnormal source processing unit is used to detect the abnormal source, and after the abnormal source has been determined by the edge side or the cloud, only the determined abnormal source is detected; when the abnormal source is not determined, the abnormal source is detected according to the matching degree order between the abnormal operation data on the edge side and the abnormal source. The cloud also includes a third communication unit, a third storage unit and an abnormal source analysis unit; the third communication unit is used to realize data interaction between the cloud and the edge side; the third storage unit is used to store historical operation data of industrial production equipment; when the abnormal source analysis unit cannot lock the abnormal source on the edge side, it uses the computing resources and stored operation data of the cloud to determine the abnormal source.

[0015] Compared with the prior art, the beneficial effects of the present invention are: reducing the delay in running data processing by the edge side, so that abnormal situations can be quickly detected by the edge side; tracing the source of abnormal situations through the collaboration between the cloud and the edge side, speeding up the tracing of the source of abnormal situations and reducing the adverse effects of abnormal situations; the edge side only sends data to the cloud when it needs to use the computing resources of the cloud to determine the source of the abnormality, rather than continuously transmitting data, which can effectively manage data flows and optimize the use of network resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a structural schematic diagram of the industrial production equipment abnormality tracing system based on cloud-edge-end collaboration of the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] Example: Figure 1 As shown, the present invention provides a technical solution, an industrial production equipment abnormality tracing system based on cloud-edge-end collaboration, including a cloud side, an edge side and a terminal side. The cloud side and the edge side are interconnected to be responsible for global operation data analysis. When the edge side cannot lock the abnormality source, the abnormality source is determined through large-scale data analysis to provide background support for abnormality tracing; the edge side and the terminal side are interconnected to perform real-time analysis and abnormality judgment on the operation data of industrial production equipment, and lock the abnormality source; the terminal side is used to collect the operation data of industrial production equipment.

[0019] The terminal side also includes a data acquisition unit and a first communication unit, the data acquisition unit is used to obtain the operation data of the industrial production equipment, and the first communication unit is used to send the obtained operation data to the edge side. The edge side also includes a second communication unit, an abnormality detection unit, a consistency determination unit, a matching degree calculation unit, a synchronization unit, a second storage unit and an abnormal source processing unit; the second communication unit is used to realize data interaction between the edge side and the cloud; the abnormality detection unit is used to perform abnormality detection on the operation data responsible for the edge side; the consistency determination unit is used to determine the consistency between the abnormal operation data detected by the edge side and the abnormal source: the matching degree calculation unit is used to determine the matching degree between the abnormal operation data detected by the edge side and the abnormal source; the synchronization unit is used to align the standard trend chart and the abnormal trend chart with the operation data collected by the edge side; the second storage unit is used to store the standard trend chart, the abnormal trend chart of the operation data and the response time data of the edge side; the abnormal source processing unit is used to detect the abnormal source, and after the abnormal source has been determined by the edge side or the cloud, only the determined abnormal source is detected; when the abnormal source is not determined, the abnormal source is detected according to the matching degree order between the abnormal operation data on the edge side and the abnormal source. The cloud also includes a third communication unit, a third storage unit and an abnormal source analysis unit; the third communication unit is used to realize data interaction between the cloud and the edge side; the third storage unit is used to store historical operation data of industrial production equipment; when the abnormal source analysis unit cannot lock the abnormal source on the edge side, it uses the computing resources and stored operation data of the cloud to determine the abnormal source.

[0020] Embodiment: The present invention provides a technical solution, a method for tracing the source of abnormalities in industrial production equipment based on cloud-edge-end collaboration, comprising: Obtain the historical operation data of industrial production equipment and establish an equipment database based on the historical operation data of industrial production equipment: Obtain historical operating data of industrial production equipment during normal operation and generate a standard trend chart for each historical operating data; Acquire historical operating data that deviates from the standard trend graph when an abnormal situation occurs in each abnormal source, establish a connection between the historical operating data that deviates from the standard trend graph and the abnormal source, and generate an abnormal trend graph for each abnormal source based on the operating data that deviates from the standard trend graph, wherein the abnormal trend graph for each abnormal source is a trend graph of part of the operating data or a trend graph of all the operating data; When an abnormal source occurs, the operation data that deviates from the standard trend graph is the relevant operation data, and the edge side responsible for the relevant operation data is the relevant edge side; for the i-th relevant edge side, obtain the response time t of other relevant edge sides, t=t i +t j , where t i is the time required for the i-th related edge side to obtain information from the cloud, t j The time required for other edge sides to send information to the cloud.

[0021] S12, the edge side obtains the operating data of industrial production equipment in real time and performs anomaly detection: Let Ak represent the kth edge side. The edge side Ak compares the acquired operation data curve with the standard trend chart of the corresponding operation data, calculates the root mean square error between the operation data and the corresponding standard trend chart, and if the root mean square error is greater than the threshold, the operation data is judged to be abnormal operation data.

[0022] The step of calculating the root mean square error between the operating data and the corresponding standard trend graph further comprises the following steps: S41, aligning the sampling nodes of the standard trend graph with the sampling nodes of the operating data curve acquired by the edge side Ak, and calculating the root mean square error between the operating data curve and the corresponding standard trend graph at the sampling nodes; S42, translate the standard trend graph by a fixed time step step, recalculate the root mean square error between the operating data curve and the corresponding standard trend graph at the sampling node of the edge side Ak, the value of the sampling time point of the standard trend graph at the edge side Ak after translation is px, and the root mean square error is calculated according to px and the value of the operating data curve at the sampling node of the edge side Ak; px=x+sign×step×v, where v represents the rate of change of the operating data on the standard trend graph, the sampling node of the edge side Ak is between the two sampling nodes of the standard trend graph after translation, and the rate of change of the operating data is obtained according to the two sampling nodes of the standard trend graph and the sampling period of the edge side Ak; when the standard trend graph is translated backward on the time axis, step is 1, and when the standard trend graph is translated forward on the time axis, sign is -1; x is the value of the operating data at the sampling node of the edge side Ak with the standard trend graph before translation; S43, each time translating a time step, when the root mean square error reaches a minimum value, determining the root mean square error between the operating data and the corresponding standard trend graph.

[0023] Identify the source of the abnormality through abnormal operation data: S31, the abnormal operation data set determined by the edge side Ak is Mk, the number of operation data contained in the abnormal operation data set Mk is mk, and from all abnormal sources, the abnormal sources that are simultaneously connected to all the operation data in the abnormal operation data set Mk are screened out, and the screened abnormal sources are set as target abnormal sources. If the number of target abnormal sources is 1, the abnormal sources are locked, and the edge side Ak processes the abnormal sources, and the process ends; if the number of target abnormal sources is not 1, the process proceeds to step S32; S32, obtaining the operation data set Na associated with the a-th target abnormal source, calculating the consistency css(k, a) of the abnormal operation data between the a-th target abnormal source and the edge side Ak, css(k, a) = (Mk∩Na) / (Mk∪Na), Mk∩Na represents the number of identical operation data in the set Mk and the set Na, and Mk∪Na represents the total number of non-repeated operation data in the set Mk and the set Na; Example: The industrial production equipment is a synchronous motor. The edge side Ak detects that the voltage at locations c1, c2, and c3 is abnormal. After the ath abnormal source appears, the voltage at locations c1, c2, c3, and c4 will be abnormal. After the bth abnormal source appears, the voltage at locations c2, c3, and c4 will be abnormal. Since the voltage at location c1 is not included after the bth abnormal source appears, the bth abnormal source is not the target abnormal source. The abnormal source is only the ath abnormal source. After the abnormal source is locked, the edge side Ak sends the information of the determined abnormal source to To the cloud, the cloud feeds back information to other edge sides to prevent other edge sides from transmitting operating data to the cloud; if the bth abnormal source appears, the voltage at locations c1, c2, c3, c4, and c5 will be abnormal, then a and b are both target abnormal sources, and the consistency css(k, a)=3 / 4, css(k, b)=3 / 5. Since the edge side Ak is not responsible for the voltage at c4 and c5, it cannot be judged and can only obtain partial information of the target abnormal source; at this time, the consistency of the target abnormal source a is higher than that of the target abnormal source b. First, the matching degree of the target abnormal source is analyzed.

[0024] S33, determine the matching degree between the a-th target abnormal source and the abnormal operation data of the edge side Ak in order of consistency from large to small, and check the abnormal source according to the matching degree from large to small; let Db represent the b-th operation data in the abnormal operation data set Mk, calculate the root mean square error between the curve of the operation data Db obtained by the edge side Ak and the abnormal trend chart of the operation data Db of the a-th target abnormal source, if the root mean square error is less than the threshold, then it is judged that the operation data Db of the edge side Ak matches the operation data Db of the a-th target abnormal source, determine the operation data set Rk that matches the a-th target abnormal source in the abnormal operation data set Mk of the edge side Ak, let the number of operation data contained in Rk be num(k, a), and then obtain the matching degree fit(k, a) between the a-th target abnormal source and the abnormal operation data of the edge side Ak, fit(k, a)=num(k, a) / (Mk∪Na).

[0025] When the edge side Ak detects that among the voltage curves at locations c1, c2, and c3, only the voltage curves at c2 and c3 match the abnormal trend graph of the target abnormal source a, then the matching degree fit(k, a)=2 / 4=1 / 2; when the edge side Ak detects that among the voltage curves at locations c1, c2, and c3, the voltage curves at c1, c2, and c3 all match the abnormal trend graph of the target abnormal source b, then the matching degree fit(k, b)=3 / 5, then before reaching the time threshold or receiving the feedback information from the cloud, the edge side Ak checks the abnormal sources in the order of b and a. Since different abnormal sources may have the same impact on the operating data when they appear, the operating data curve at the edge side detection can be matched with the abnormal trend graphs of multiple abnormal sources at the same time.

[0026] The calculation of the root mean square error between the curve of the operating data Db acquired by the edge side Ak and the abnormal trend graph of the operating data Db of the ath target abnormal source also includes the following steps: According to steps S41, S42 and S43, the abnormal trend graph of the operating data Db of the a-th target abnormal source is translated and the root mean square error after each translation is calculated. When the root mean square error reaches the minimum value, the root mean square error between the curve of the operating data Db obtained by the edge side Ak and the abnormal trend graph of the operating data Db of the a-th target abnormal source is determined.

[0027] If the edge side cannot lock the source of the exception, the operating data of the industrial production equipment will be transmitted to the cloud, and the process goes to step S13; if the edge side can lock the source of the exception, the operating data of the industrial production equipment will not be transmitted to the cloud, and the edge side will process the source of the exception and the process ends.

[0028] In step S12, the following steps are also included: When the edge side Ak determines that there is abnormal operation data through judgment, if the edge side Ak can lock the abnormal source, the edge side Ak sends the information of the determined abnormal source to the cloud. After the cloud receives the information of the determined abnormal source, it feeds back the information of the determined abnormal source to all edge sides; if the edge side Ak does not lock the abnormal source, and within the time threshold, does not receive the information of the determined abnormal source sent by the cloud, then the edge side Ak transmits the operation data of the industrial production equipment to the cloud, and the cloud determines the abnormal source, and at the same time transmits the consistency and matching degree of the abnormal operation data of the edge side Ak with all target abnormal sources to the cloud; the time threshold is determined by the following steps: the edge side Ak obtains the response time of the relevant edge side of the ath target abnormal source, takes the maximum value of all response times of the ath target abnormal source to obtain the response time of the ath target abnormal source; takes the maximum value of the response time of all target abnormal sources to obtain the time threshold.

[0029] S13: The cloud analyzes the abnormal sources according to the priority based on the operation data transmitted by the edge side: The cloud determines the priority of the abnormal source based on the consistency and matching degree of the abnormal operation data received from each edge side and all target abnormal sources. For the ath target abnormal source, the priority );W1 and W2 are weights; After determining the priorities of all target abnormal sources, the cloud analyzes the abnormal sources based on the operation data transmitted by the edge side in descending order of priority.

[0030] Continuing with the case where both a and b are target abnormal sources, when the edge side Ah is responsible for the voltages at c4 and c5, and detects that the voltage at c4 is abnormal, then , ; When the voltage curves at c4 match the abnormal trend graphs of the target abnormal sources a and b, , ,Since no abnormal situation is detected in the voltage at c5, it does not match the abnormal trend graph of the target abnormal source b; and Can be set according to needs, such as Take it as 0.4, Take it as 0.6; calculate , , check a and b one by one according to their priority order.

[0031] After the abnormal source is determined, the abnormal source information is fed back to the corresponding edge side, and the edge side processes the abnormal source.

[0032] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. The method for tracing the abnormality of industrial production equipment based on cloud-edge-end collaboration is characterized by: The following steps are involved: S11, acquiring historical operation data of industrial production equipment, and establishing an equipment database based on the historical operation data of the industrial production equipment; S12, the edge side obtains the operating data of the industrial production equipment in real time and performs anomaly detection, and locks the abnormal source through the abnormal operating data. If the edge side cannot lock the abnormal source, the operating data of the industrial production equipment is transmitted to the cloud, and the process goes to step S13; If the edge side can lock the abnormal source, the operation data of the industrial production equipment will not be transmitted to the cloud, and the edge side will process the abnormal source and end; S13, the cloud analyzes the abnormal source according to the priority based on the operation data transmitted by the edge side, and after determining the abnormal source, feeds back the abnormal source information to the corresponding edge side, and the edge side processes the abnormal source and ends.

2. The method for tracing the abnormality of industrial production equipment based on cloud-edge-end collaboration according to claim 1 is characterized in that: In step S11, the establishment of the equipment database based on the historical operation data of the industrial production equipment further includes the following steps: S21, obtaining historical operation data of the industrial production equipment when it is working normally and generating a standard trend graph of each historical operation data; S22, obtaining historical operating data of each abnormal source that deviates from the standard trend graph when an abnormal situation occurs, establishing a connection between the historical operating data that deviates from the standard trend graph and the abnormal source, and generating an abnormal trend graph of each abnormal source based on the operating data that deviates from the standard trend graph, wherein the abnormal trend graph of each abnormal source is a trend graph of part of the operating data or a trend graph of all the operating data; S23, when an abnormal source occurs, the operation data that deviates from the standard trend graph is defined as the relevant operation data, and the edge side responsible for the relevant operation data is defined as the relevant edge side; for the i-th relevant edge side, the response time t of other relevant edge sides is obtained, t=t i +t j , where t i is the time required for the i-th related edge side to obtain information from the cloud, t j The time required for other edge sides to send information to the cloud.

3. The method for tracing the abnormality of industrial production equipment based on cloud-edge-end collaboration according to claim 1 is characterized in that: In step S12, the edge side acquires the operation data of the industrial production equipment in real time and performs anomaly detection, and further includes the following steps: S30, let Ak represent the kth edge side, the edge side Ak compares the acquired operation data curve with the standard trend graph of the corresponding operation data, calculates the root mean square error between the operation data and the corresponding standard trend graph, and if the root mean square error is greater than a threshold, determines that the operation data is abnormal operation data; The method of locking the abnormal source through abnormal operation data comprises the following steps: S31, the abnormal operation data set determined by the edge side Ak is Mk, the number of operation data contained in the abnormal operation data set Mk is mk, and from all abnormal sources, the abnormal sources that are simultaneously connected to all the operation data in the abnormal operation data set Mk are screened out, and the screened abnormal sources are set as target abnormal sources. If the number of target abnormal sources is 1, the abnormal sources are locked, and the edge side Ak processes the abnormal sources, and the process ends; if the number of target abnormal sources is not 1, the process proceeds to step S32; S32, obtaining the operation data set Na associated with the a-th target abnormal source, calculating the consistency css(k, a) of the abnormal operation data between the a-th target abnormal source and the edge side Ak, css(k, a) = (Mk∩Na) / (Mk∪Na), Mk∩Na represents the number of identical operation data in the set Mk and the set Na, and Mk∪Na represents the total number of non-repeated operation data in the set Mk and the set Na; S33, determine the matching degree between the a-th target abnormal source and the abnormal operation data of the edge side Ak in order of consistency from large to small, and check the abnormal source according to the matching degree from large to small; let Db represent the b-th operation data in the abnormal operation data set Mk, calculate the root mean square error between the curve of the operation data Db obtained by the edge side Ak and the abnormal trend chart of the operation data Db of the a-th target abnormal source, if the root mean square error is less than the threshold, then it is judged that the operation data Db of the edge side Ak matches the operation data Db of the a-th target abnormal source, determine the operation data set Rk that matches the a-th target abnormal source in the abnormal operation data set Mk of the edge side Ak, let the number of operation data contained in Rk be num(k, a), and then obtain the matching degree fit(k, a) between the a-th target abnormal source and the abnormal operation data of the edge side Ak, fit(k, a)=num(k, a) / (Mk∪Na).

4. The method for tracing the abnormality of industrial production equipment based on cloud-edge-end collaboration according to claim 3 is characterized in that: In step S30, the step of calculating the root mean square error between the operating data and the corresponding standard trend graph further includes the following steps: S41, aligning the sampling nodes of the standard trend graph with the sampling nodes of the operating data curve acquired by the edge side Ak, and calculating the root mean square error between the operating data curve and the corresponding standard trend graph at the sampling nodes; S42, translate the standard trend graph by a fixed time step step, recalculate the root mean square error between the operating data curve and the corresponding standard trend graph at the sampling node of the edge side Ak, the value of the sampling time point of the standard trend graph at the edge side Ak after translation is px, and the root mean square error is calculated according to px and the value of the operating data curve at the sampling node of the edge side Ak; px=x+sign×step×v, where v represents the rate of change of the operating data on the standard trend graph, the sampling node of the edge side Ak is between the two sampling nodes of the standard trend graph after translation, and the rate of change of the operating data is obtained according to the two sampling nodes of the standard trend graph and the sampling period of the edge side Ak; when the standard trend graph is translated backward on the time axis, step is 1, and when the standard trend graph is translated forward on the time axis, sign is -1; x is the value of the operating data at the sampling node of the edge side Ak with the standard trend graph before translation; S43, each time translating a time step, when the root mean square error reaches a minimum value, determining the root mean square error between the operating data and the corresponding standard trend graph; In step S33, the calculation of the root mean square error between the curve of the operating data Db obtained by the edge side Ak and the abnormal trend graph of the operating data Db of the a-th target abnormal source also includes the following steps: translating the abnormal trend graph of the operating data Db of the a-th target abnormal source in accordance with steps S41, S42 and S43 and calculating the root mean square error after each translation; when the root mean square error reaches the minimum value, determining the root mean square error between the curve of the operating data Db obtained by the edge side Ak and the abnormal trend graph of the operating data Db of the a-th target abnormal source.

5. The method for tracing the abnormality of industrial production equipment based on cloud-edge-end collaboration according to claim 4 is characterized in that: In step S12, the following steps are also included: When the edge side Ak determines that there is abnormal operation data through judgment, if the edge side Ak can lock the abnormal source, the edge side Ak sends the information of the determined abnormal source to the cloud. After the cloud receives the information of the determined abnormal source, it feeds back the information of the determined abnormal source to all edge sides; if the edge side Ak does not lock the abnormal source, and within the time threshold, does not receive the information of the determined abnormal source sent by the cloud, then the edge side Ak transmits the operation data of the industrial production equipment to the cloud, and the cloud determines the abnormal source, and at the same time transmits the consistency and matching degree of the abnormal operation data of the edge side Ak with all target abnormal sources to the cloud; the time threshold is determined by the following steps: the edge side Ak obtains the response time of the relevant edge side of the ath target abnormal source, takes the maximum value of all response times of the ath target abnormal source to obtain the response time of the ath target abnormal source; takes the maximum value of the response time of all target abnormal sources to obtain the time threshold.

6. The method for tracing the abnormality of industrial production equipment based on cloud-edge-end collaboration according to claim 5 is characterized in that: In step S13, analyzing the abnormal source according to the priority also includes the following steps: The cloud determines the priority of the abnormal source based on the consistency and matching degree of the abnormal operation data received from each edge side and all target abnormal sources. For the ath target abnormal source, the priority ;W1 and W2 are weights;After determining the priorities of all target abnormal sources, the cloud analyzes the abnormal sources based on the operation data transmitted from the edge side in descending order of priority.

7. An industrial production equipment anomaly tracing system based on cloud-edge-end collaboration that adopts the industrial production equipment anomaly tracing method as described in any one of claims 1 to 6, characterized in that: It includes a cloud side, an edge side and a terminal side. The cloud side is interconnected with the edge side and is responsible for global operation data analysis. When the edge side cannot lock the abnormal source, the abnormal source is determined through large-scale data analysis to provide background support for abnormal tracing; the edge side is interconnected with the terminal side and is used to perform real-time analysis and abnormal judgment on the operation data of industrial production equipment, and lock the abnormal source; the terminal side is used to collect the operation data of industrial production equipment.

8. The industrial production equipment abnormality tracing system based on cloud-edge-end collaboration according to claim 7 is characterized in that: The terminal side also includes a data acquisition unit and a first communication unit. The data acquisition unit is used to acquire the operation data of the industrial production equipment, and the first communication unit is used to send the acquired operation data to the edge side.

9. The industrial production equipment abnormality tracing system based on cloud-edge-end collaboration according to claim 8 is characterized in that: The edge side also includes a second communication unit, an anomaly detection unit, a consistency determination unit, a matching degree calculation unit, a synchronization unit, a second storage unit and an abnormal source processing unit; the second communication unit is used to realize data interaction between the edge side and the cloud; the anomaly detection unit is used to perform anomaly detection on the operation data responsible for the edge side; the consistency determination unit is used to determine the consistency between the abnormal operation data detected by the edge side and the abnormal source: the matching degree calculation unit is used to determine the matching degree between the abnormal operation data detected by the edge side and the abnormal source; The synchronization unit is used to align the standard trend graph and the abnormal trend graph with the operation data collected by the edge side; The second storage unit is used to store a standard trend graph, an abnormal trend graph and edge-side response time data of the operation data; The abnormal source processing unit is used to detect the abnormal source. After the abnormal source has been determined by the edge side or the cloud, only the determined abnormal source is detected; When the abnormal source is not determined, the abnormal source is detected according to the matching order between the abnormal operation data on the edge side and the abnormal source.

10. The industrial production equipment abnormality tracing system based on cloud-edge-end collaboration according to claim 9 is characterized in that: The cloud also includes a third communication unit, a third storage unit and an abnormal source analysis unit; the third communication unit is used to realize data interaction between the cloud and the edge side; the third storage unit is used to store historical operation data of industrial production equipment; when the abnormal source analysis unit cannot lock the abnormal source on the edge side, it uses the computing resources and stored operation data of the cloud to determine the abnormal source.

Images