An edge computing gateway

Abstract

This invention discloses an edge computing gateway applied in the gateway field, comprising a gateway hardware layer and a system layer. The system layer includes a device fingerprint database, a data receiving module, a processing module, a risk assessment module, an index reference database, an instruction database, and a fault preliminary judgment module. By presetting multiple data reference ranges for each monitoring parameter and assigning different risk indices and emergency instructions to each data reference range, different levels of emergency measures are taken for the device based on the risk level of the data, thus controlling the operational risks of the device in a timely manner. When the device data is abnormal, by combining the temperature of the gateway hardware layer and other device statuses, it can effectively distinguish between three types of faults: the device layer, the gateway hardware layer, and the data transmission layer, providing maintenance personnel with a clear direction for troubleshooting and significantly reducing the average repair time. In addition, by grouping devices according to different related factors, it can quickly determine whether the data anomaly is caused by shared related factors, improving the diagnostic speed.

Description

Technical Field

[0001] This invention relates to the field of gateways, and more particularly to an edge computing gateway. Background Technology

[0002] Edge gateways are devices that connect terminal devices and cloud servers, playing a crucial role in IoT systems. They act as an intermediary layer between terminal devices and cloud servers, responsible for collecting data from terminal devices, processing and storing it, and sending the processed and packaged data to the cloud server. By connecting to IoT devices, edge gateways can achieve real-time control and monitoring of devices, and analyze and make decisions based on the data. Simultaneously, edge gateways can coordinate communication between terminal devices, thereby enabling effective management and control of the entire IoT environment.

[0003] For example, Chinese patent application CN118827419A discloses a management and control system for IoT gateways, which can quickly identify anomalies and also has risk index calculation and fault prediction capabilities, effectively solving the problem of complex fault diagnosis in traditional IoT gateways and facilitating the management and control of IoT gateways. Another example is Chinese patent application CN119676097A, which discloses a remote management and control method for smart gateway devices based on the HarmonyOS system. This method achieves tight connection and data sharing between the gateway device and the system, collecting real-time operating status data of the gateway and its connected sensors through a unified device management platform. Combined with machine learning algorithms, a fault prediction model is trained. This model can predict the potential fault risks of the gateway device, including the type, level, and time point of the fault. The trained model is deployed on the edge computing node of the HarmonyOS system to analyze and process data in real time, predict device faults, and automatically generate control commands according to preset strategies. The commands are sent to the gateway device immediately, realizing remote management and control.

[0004] When edge computing gateways collect and process data, they are prone to data anomalies due to various factors (such as environmental factors, abnormal data transmission networks, gateway malfunctions, and edge device malfunctions). In existing technologies, the gateway typically reports the abnormal data to the cloud control center for judgment. This approach has high response latency and cannot meet the requirements for real-time fault handling. Summary of the Invention

[0005] The core of this invention lies in constructing a multi-dimensional fault type judgment system within the gateway to achieve preliminary root cause analysis of abnormal device data. Based on the analysis results, differentiated emergency handling strategies are matched, solving the problems of high fault response delay, untimely real-time processing requirements, and unclear troubleshooting direction in existing technologies. At the same time, by grouping devices according to different related factors, it is possible to quickly determine whether data anomalies are caused by shared factors, simplifying the fault diagnosis process and improving the diagnosis speed.

[0006] To solve the above problems, the present invention adopts the following technical solution.

[0007] An edge computing gateway includes a gateway hardware layer and a system layer. The system layer includes a device fingerprint database, a data receiving module, a processing module, a risk assessment module, an index reference database, an instruction database, and a fault preliminary assessment module. The device fingerprint database stores static fingerprints, including the MAC addresses of multiple devices. The index reference database stores multiple preset data reference ranges for each device and a risk index corresponding to each data reference range, and the multiple risk indices are different. The risk assessment module is used to compare the device's real-time data with the data reference ranges to determine its risk index. The instruction database stores emergency instructions that correspond one-to-one with the multiple risk indices. When the device's real-time data is not within any preset data reference range, the fault preliminary assessment module determines the fault type, which includes device layer faults, gateway hardware layer faults, and data transmission layer faults.

[0008] An edge computing gateway, the method of using which includes the following steps:

[0009] S1. The system layer continuously receives real-time data from various devices, processes the real-time data, and then transmits it to the control center. At the same time, it compares the real-time data with the data in the index reference library.

[0010] S2. When the real-time data of device A meets a certain data reference range, select the emergency command corresponding to the risk index according to the risk index corresponding to the data reference range, and send it to device A.

[0011] S3. When the real-time data of device A is outside any of its corresponding data reference range, the real-time data status is recorded as abnormal, and the following multi-dimensional fault diagnosis operations are performed:

[0012] S3-1. Obtain the current temperature of the gateway hardware layer and the real-time data status of other devices at the same time.

[0013] S3-2, Scenario 1: The temperature of the gateway hardware layer is normal, but the real-time data of some other devices is abnormal. It is initially determined that there is a fault in the data transmission layer between the gateway hardware layer and the devices.

[0014] S3-3, Scenario 2: The gateway hardware layer temperature is normal, and the real-time data of other devices are also normal. The initial judgment is that device A is faulty.

[0015] S3-4, Scenario 3: Abnormal temperature of the gateway hardware layer and abnormal real-time data of some other devices. It is initially determined to be a fault in the gateway hardware layer or a fault in the data transmission layer. Then, the cooling of the gateway hardware layer is increased, and real-time data of all devices is received for a second judgment. When the real-time data of a certain device recovers from the previous abnormality to normal, or changes from normal to abnormal, it is determined to be a fault in the gateway hardware layer. When the real-time data status of the device remains unchanged, it is determined to be a fault in the data transmission layer.

[0016] S3-5, Scenario 4: The gateway hardware layer temperature is abnormal, while the real-time data of other devices are normal. It is initially determined that the gateway hardware layer or device A is faulty. Then, the cooling of the gateway hardware layer is increased, and the real-time data of all devices is received for a second judgment. If the real-time data of device A is still abnormal, and the data of other devices are normal, then device A is determined to be faulty. If the real-time data of some other devices is also abnormal, then the gateway hardware layer is determined to be faulty.

[0017] S4. When device A is determined to be faulty, perform the following operations:

[0018] S4-1. Similarity Analysis: Analyze the similarity between the real-time data of device A and all preset data reference ranges, and find the data reference range Q with the greatest similarity.

[0019] S4-2, Command Selection and Issuance: Select a data reference range W with a risk index higher than the risk index corresponding to the data range Q of the device, and ensure that the risk indices corresponding to the data reference range W and the data reference range Q are adjacent. Then, retrieve the emergency command corresponding to the risk index of the data reference range W from the command library and issue it to the device A.

[0020] Furthermore, when a gateway hardware layer fault is detected, a gateway warning is sent to the control center; when a data transmission layer fault is detected, a network warning is sent to the control center, and data transmission to the control center is suspended.

[0021] Optionally, the system layer also includes a pre-variant database, which stores multiple device pre-variant data ranges that correspond one-to-one with emergency commands. In step S4-2, after the system layer sends an emergency command to device A, it continues to receive real-time data from device A. If the real-time data of device A does not conform to the device pre-variant data range corresponding to the emergency command within a set time, it is determined that the device is out of control. The system layer issues an emergency warning to the control center. At the same time, the system layer retrieves the emergency command with the highest risk index from the command library and sends it to device A.

[0022] Optionally, the device fingerprint database also stores dynamic fingerprints, which include historical behavior data of multiple devices. The historical behavior data includes data evolution trends, data anomaly cycles, and data anomaly moments.

[0023] Optionally, in step S3, when the real-time data of device A is not within any of its corresponding data reference range, it is first determined whether this abnormal data of device A conforms to historical behavior data. If it does not conform, then a multi-dimensional fault diagnosis operation is performed. If it conforms, no other operation is required.

[0024] Optionally, the system layer also includes a device association database, which stores multiple device association groups. Each device association group consists of multiple MAC addresses with association factors, including the same environment, the same physical space, and the same target.

[0025] Optionally, in step S3-1, when the real-time data of some other devices also shows abnormalities, the following operations are performed first: determine whether the MAC address of this part of the devices is associated with the MAC address of device A. If the MAC address of this part of the devices and the MAC address of device A belong to the same device association group, then the abnormality is determined to be environmental abnormality, spatial abnormality, or object abnormality based on the association factors of the device association group. Otherwise, if the MAC address of this part of the devices and the MAC address of device A do not belong to any device association group, then proceed to step S3-2 or step S3-4.

[0026] Compared with the prior art, the advantages of this invention are:

[0027] (1) This solution presets multiple data reference ranges for each monitoring parameter and assigns different risk indices and corresponding emergency instructions to each data reference range. Based on the risk level of the data, different levels of emergency measures are taken for the equipment, thus realizing the gradient management of risk data. When the equipment data is abnormal, combined with the temperature of the gateway hardware layer and the status of other equipment, it can effectively distinguish between three types of faults: equipment layer, gateway hardware layer, and data transmission layer. This provides maintenance personnel with a clear direction for troubleshooting and significantly reduces the average repair time. When the equipment is abnormal, an emergency instruction of a higher risk level is issued to the equipment to control the risk of equipment operation in a timely manner.

[0028] (2) After issuing an emergency command to the equipment that has malfunctioned, the system continuously monitors whether the equipment data has entered the expected range of pre-variable data, forming a closed-loop operation of command issuance and effect verification. When the equipment is out of control or the actuator fails, the system can detect it in time and initiate the emergency command with the highest risk index. As an effective safety redundancy design, it further improves the safety of the equipment.

[0029] (3) Using historical behavioral data such as data evolution trends, data anomaly cycles and data anomaly times to identify the authenticity of equipment data anomalies can effectively reduce unnecessary fault diagnosis operations and improve system efficiency.

[0030] (4) By setting up a device association database, devices can be grouped based on environment, physical space or target object. When multiple devices malfunction at the same time, it can quickly determine whether the data abnormality is caused by shared factors (such as environmental changes or carrier failure), which simplifies the fault diagnosis process and improves the diagnosis speed. It is especially suitable for collaborative fault analysis in complex industrial scenarios. Attached Figure Description

[0031] Figure 1 The system of the present invention Figure 1 ;

[0032] Figure 2 This is a schematic diagram illustrating the correspondence between the index reference library and the instruction library of the present invention;

[0033] Figure 3 This is a schematic diagram of the device fingerprint database of the present invention;

[0034] Figure 4 The process of this invention Figure 1 ;

[0035] Figure 5 This is a schematic diagram of the structure of the device association database of the present invention;

[0036] Figure 6 The process of this invention Figure 2 ; Detailed Implementation

[0037] The technical solutions will now be clearly and completely described with reference to the accompanying drawings in the embodiments of the present invention.

[0038] First implementation method:

[0039] Please see Figure 1 An edge computing gateway includes a gateway hardware layer and a system layer. The system layer includes a device fingerprint database, a data receiving module, a processing module, a risk assessment module, an index reference database, an instruction database, and a fault preliminary assessment module. The device fingerprint database stores static fingerprints, which include the MAC addresses of multiple devices. Under normal circumstances, the MAC address can serve as a unique identifier for the device, facilitating rapid connection between the device and the edge computing gateway. The devices connected to the gateway can be various industrial equipment or various sensors. The data receiving module is used to receive actual data collected by multiple devices, and the processing module is used to process and analyze the received data before transmitting the processed data to the control center.

[0040] The index reference library stores multiple preset data reference ranges for each device and the risk index corresponding to each data reference range. The multiple risk indices are different. The risk judgment module is used to compare the device's real-time data with the data reference range to determine its risk index. The instruction library stores emergency instructions that correspond one-to-one with the multiple risk indices. When the device's real-time data is not within any preset data reference range, the fault initial judgment module judges the fault type. The fault types include device layer faults, gateway hardware layer faults, and data transmission layer faults.

[0041] Specifically, let's assume a device connected to this edge computing gateway is a temperature monitoring device. In this embodiment, n data reference ranges are preset for the temperature monitoring device, namely T1-T2 (normal temperature), T2-T3 (slightly high temperature), T3-T4 (moderate high temperature)...Tn-Tm (extreme high temperature). The risk indices corresponding to the n data reference ranges are R1 (no risk), R2 (very low risk), R3 (moderate risk)...Rn (extreme risk), and the instructions corresponding to the multiple risk indices are L1 (no action required, normal monitoring), L2 (increase cooling power), L3 (decrease sampling frequency)...Ln (device shutdown). Under normal circumstances, the real-time data received by the temperature monitoring device from the system layer should be within a certain data reference range. At this time, the system layer can select the corresponding emergency instruction and send it to the temperature sensor according to the correspondence of the risk indices, so that the device can perform emergency operations in time when the data is abnormal, effectively controlling the operational risk of the device. Of course, under abnormal circumstances, the real-time data received by the temperature monitoring device from the system layer may not be within a certain data reference range. In this case, the abnormal situation can be specifically analyzed (refer to the following step S3 for details).

[0042] An edge computing gateway, the method of using which includes the following steps:

[0043] S1. The system layer continuously receives real-time data from various devices, processes the real-time data, and then transmits it to the control center. At the same time, it compares the real-time data with the data in the index reference library.

[0044] S2. When the real-time data of device A meets a certain data reference range, select the emergency command corresponding to the risk index according to the risk index corresponding to the data reference range, and send it to device A.

[0045] S3. When the real-time data of device A is outside any of its corresponding data reference range, the real-time data status is recorded as abnormal, and the following multi-dimensional fault diagnosis operations are performed:

[0046] S3-1. Obtain the current temperature of the gateway hardware layer and the real-time data status of other devices at the same time.

[0047] S3-2, Scenario 1: The temperature of the gateway hardware layer is normal, but the real-time data of some other devices is abnormal. It is initially determined that there is a fault in the data transmission layer between the gateway hardware layer and the devices.

[0048] Since the probability of multiple devices failing simultaneously is low, and the normal temperature of the gateway hardware layer largely reflects the normal state of the gateway itself, the data anomaly is initially determined to be caused by an abnormality in the network transmission link (i.e., the data transmission layer).

[0049] S3-3, Scenario 2: The gateway hardware layer temperature is normal, and the real-time data of other devices are also normal. The initial judgment is that device A is faulty.

[0050] S3-4, Scenario 3: Abnormal temperature of the gateway hardware layer and abnormal real-time data of some other devices. It is initially determined to be a fault in the gateway hardware layer or a fault in the data transmission layer. Then, the cooling of the gateway hardware layer is increased, and real-time data of all devices is received for a second judgment. When the real-time data of a certain device recovers from the previous abnormality to normal, or changes from normal to abnormal, it is determined to be a fault in the gateway hardware layer. When the real-time data status of the device remains unchanged, it is determined to be a fault in the data transmission layer.

[0051] Gateway hardware layers (such as the CPU) are prone to instability at high temperatures, leading to data processing errors. By increasing the cooling rate (e.g., increasing fan speed), the temperature of the gateway hardware layer can be lowered more quickly, which can verify whether high temperature is the direct cause of data anomalies. During the temperature drop, data continues to be received and processed. If there are sudden changes between normal and abnormal data in subsequent received data, it indicates that the gateway itself is unstable, and thus it is determined to be a gateway hardware layer failure. Conversely, if the data status of the device remains unchanged, it indicates that the gateway itself is operating stably, and it is determined to be a data transmission layer failure.

[0052] S3-5, Scenario 4: The gateway hardware layer temperature is abnormal, while the real-time data of other devices are normal. The initial judgment is that the gateway hardware layer or device A is faulty. Then, the cooling of the gateway hardware layer is increased, and the real-time data of all devices is received for a second judgment. If the real-time data of device A is still abnormal, and the data of other devices are normal, then device A is judged to be faulty. If the real-time data of some other devices is also abnormal, it indicates that the scope of the fault is expanding or unstable, then the gateway hardware layer is judged to be faulty.

[0053] S4. When device A is determined to be faulty, perform the following operations:

[0054] S4-1. Similarity Analysis: Analyze the similarity between the real-time data of device A and all preset data reference ranges, and find the data reference range Q with the greatest similarity.

[0055] S4-2, Command Selection and Issuance: Select a data reference range W with a risk index higher than the risk index corresponding to the data range Q of the device, and ensure that the risk indices corresponding to the data reference range W and the data reference range Q are adjacent. Then, retrieve the emergency command corresponding to the risk index of the data reference range W from the command library and issue it to the device A.

[0056] Through the similarity analysis and risk escalation strategy selected in step S4, an emergency command slightly higher than the current risk assessment level is issued. This makes the emergency command more adaptable, more effective in controlling equipment risks, and also reduces the problems in normal equipment use caused by overly aggressive commands.

[0057] In step S3, when a preliminary judgment result of a gateway hardware layer failure is obtained, a gateway warning is issued to the control center; when a preliminary judgment result of a data transmission layer failure is obtained, a network warning is issued to the control center, and data transmission to the control center is suspended, so that external risks are not easily spread to the control center through data transmission.

[0058] This application achieves gradient management of risk data by pre-setting multiple data reference ranges for each monitoring parameter and assigning different risk indices and corresponding emergency instructions to each data reference range. Based on the risk level of the data, different levels of emergency measures are taken for the equipment. When the equipment data is abnormal, combined with the temperature of the gateway hardware layer and the status of other equipment, it can effectively distinguish between three types of faults: equipment layer, gateway hardware layer, and data transmission layer. This provides maintenance personnel with a clear direction for troubleshooting and significantly reduces the average repair time. When the equipment is abnormal, an emergency instruction of a higher risk level is issued to the equipment to control the risk of equipment operation in a timely manner.

[0059] Second implementation method:

[0060] Based on the first embodiment, this implementation adds a pre-variant database to the system layer. The pre-variant database stores multiple device pre-variant data ranges that correspond one-to-one with emergency commands. In step S4-2, after the system layer sends an emergency command to device A, it continues to receive real-time data from device A. If the real-time data of device A does not conform to the device pre-variant data range corresponding to the emergency command within a set time, it is determined that the device is out of control. The system layer issues an emergency warning to the control center. At the same time, the system layer retrieves the emergency command with the highest risk index from the command library and sends it to device A.

[0061] Under normal circumstances, after device A receives and executes an emergency command, the data subsequently collected by device A should conform to the expected pre-varying data range, indicating that device A has performed a normal emergency handling operation. This is a verification step to determine the controllability of device A. Since it takes a certain amount of time for the command to reach the device and produce an effect, the "set time" can be preset according to the device characteristics and command type. If the emergency command fails to produce the expected effect within the set time, it is determined that the device is out of control, indicating that device A may have a complete failure, actuator failure, or other similar situations. At this time, an emergency warning is issued to the control center to facilitate timely handling by personnel. At the same time, a redundant operation is performed: the emergency command with the highest risk index is issued to device A to further extend the controllability of device A. Generally, the emergency command with the highest risk index is the device shutdown command. If device A executes this command normally, it can further improve the safety of device A.

[0062] The third implementation method:

[0063] This embodiment adds the following to the first or second embodiment: The device fingerprint database also stores dynamic fingerprints, which include historical behavior data of multiple devices. The historical behavior data includes data evolution trends, data anomaly cycles, and data anomaly moments. The data evolution trend is the changing state of the device within a continuous time period, such as continuous rise, continuous fall, wave-like alternation of rise and fall, etc. The data anomaly cycle is the cyclical period in which the device experiences data mutations. The data after the mutation may not be within the data reference range, but since the data mutation is a periodic regular change, it can be classified as a normal data change of the device. The data anomaly moment is a preset time point when the device experiences data anomalies, such as the time of normal device maintenance. Since it is an expected anomaly, it is classified as a normal state.

[0064] By introducing the aforementioned historical behavior data, a quick means of judging the authenticity of abnormal data can be provided. Specifically: in step S3, when the real-time data of device A is not within any of its corresponding data reference range, it is first determined whether this abnormal data of device A conforms to the historical behavior data. If it does not conform, then a multi-dimensional fault diagnosis operation is performed. If it conforms, no other operation is required, that is, no multi-dimensional fault diagnosis operation is required, and this data abnormality is a data change under the expected situation.

[0065] Fourth implementation method:

[0066] Based on the first, second, or third implementation, this embodiment adds the following: The system layer also includes a device association database, which stores multiple device association groups. Each device association group consists of multiple MAC addresses with association factors, including but not limited to the same environment, the same physical space, and the same target.

[0067] The principles for setting related factors are as follows: "Same environment" can refer to the same installation environment or the same usage environment. When two or more devices are installed and used in the same environment, they will be affected by the same environmental factors (including ambient temperature, humidity, etc.), resulting in synchronized data anomalies. "Same physical space" can refer to two or more devices installed on the same carrier, such as the same platform or the same pipeline, so that even if multiple devices are not in the same environment, they are affected by the same carrier. For example, when the carrier experiences severe vibration, the vibration will be transmitted to multiple devices simultaneously, causing abnormal changes in their monitoring data. "Same object of action" can refer to two or more devices collecting data on the same object. For example, when multiple devices monitor the same transformer, and the transformer stops due to a fault, multiple devices will simultaneously detect sudden data changes (e.g., vibration sensors cannot detect equipment vibration data, and current sensors cannot detect current data).

[0068] Therefore, by setting correlation factors, the monitoring status of multiple devices can be linked to form multiple different device association groups. When the monitoring data of a device is abnormal, it is possible to quickly determine certain abnormal situations, as follows:

[0069] In step S3-1, when the real-time data of some other devices also shows anomalies, the following operations are performed first: determine whether the MAC address of these devices is associated with the MAC address of device A. If the MAC address of these devices and the MAC address of device A belong to the same device association group, then the association factors of the device association group are used to determine whether the problem is due to environmental anomalies, spatial anomalies, or anomalies of the target object. Otherwise, if the MAC address of these devices and the MAC address of device A do not belong to any device association group, then proceed to step S3-2 or step S3-4.

[0070] The above description is merely a preferred embodiment of the present invention; it encompasses all the protection scope of the present invention. Any equivalent substitutions or modifications made by those skilled in the art within the technical scope disclosed in the present invention, based on the technical solutions and improved concepts of the present invention, should be covered within the protection scope of the present invention.

Claims

1. An edge computing gateway, characterized by: The gateway comprises a gateway hardware layer and a system layer, the system layer comprises a device fingerprint library, a data receiving module, a processing module, a risk judgment module, an index reference library, an instruction library and a fault preliminary judgment module, the device fingerprint library stores static fingerprints, the static fingerprints comprise MAC addresses of multiple devices, the index reference library stores multiple data reference ranges preset for each device and risk indexes corresponding to each data reference range, the multiple risk indexes are different, the risk judgment module is used for comparing real-time data of a device with the data reference ranges to judge a risk index, the instruction library stores emergency instructions corresponding to the multiple risk indexes one by one, when the real-time data of the device is not in any preset data reference range, the fault preliminary judgment module judges a fault type, the fault type comprises a device layer fault, a gateway hardware layer fault and a data transmission layer fault; The use method of the edge computing gateway comprises the following steps: S1, the system layer continuously receives real-time data from each device, processes the real-time data and then transmits the real-time data to a control center, and meanwhile, compares the real-time data with data in the index reference library; S2, when the real-time data of a device A meets a certain data reference range, according to a risk index corresponding to the data reference range, an emergency instruction corresponding to the risk index is selected and sent to the device A; S3, when the real-time data of the device A is not in any data reference range corresponding to the device A, the real-time data state is recorded as abnormal, and the following multi-dimensional fault diagnosis operation is performed: S3-1, the current temperature of the gateway hardware layer and the real-time data states of other devices at the same time are acquired; S3-2, case one: the temperature of the gateway hardware layer is normal, and the real-time data of part of the other devices is abnormal, it is preliminarily judged that there is a fault in the data transmission layer between the gateway hardware layer and the devices; S3-3, case two: the temperature of the gateway hardware layer is normal, and the real-time data of the other devices is normal, it is preliminarily judged that the device A is faulty; S3-4, case three: the temperature of the gateway hardware layer is abnormal, and the real-time data of part of the other devices is abnormal, it is preliminarily judged that the gateway hardware layer is faulty or the data transmission layer is faulty, then the cooling amplitude of the gateway hardware layer is increased, the real-time data of all the devices is continuously received, and secondary judgment is performed, when the real-time data of a certain device is restored to normal from abnormal in the previous stage or changes from normal to abnormal, it is judged that the gateway hardware layer is faulty, when the real-time data states of the devices are unchanged, it is judged that the data transmission layer is faulty; S3-5, case four: the temperature of the gateway hardware layer is abnormal, and the real-time data of the other devices is normal, it is preliminarily judged that the gateway hardware layer is faulty or the device A is faulty, then the cooling amplitude of the gateway hardware layer is increased, the real-time data of all the devices is continuously received, and secondary judgment is performed, when the real-time data of the device A is still abnormal and the real-time data of the other devices is normal, it is judged that the device A is faulty, when the real-time data of part of the other devices is also abnormal, it is judged that the gateway hardware layer is faulty; The system layer further comprises a device association database, the device association database stores a plurality of device association groups, each device association group is composed of a plurality of MAC addresses with association factors, the association factors include the same environment, the same physical space and the same action object, when the real-time data of part of other devices also exist abnormities in step S3-1, the following operations are performed first: whether the MAC address of the part of devices is associated with the MAC address of device A, when the MAC address of the part of devices and the MAC address of device A belong to the same device association group, it is judged as environment abnormality, space abnormality or action object abnormality according to the association factors of the device association group, otherwise, when the MAC address of the part of devices and the MAC address of device A do not belong to any device association group, step S3-2 or step S3-4 is performed again; S4, when it is determined that device A is faulty, the following operations are performed: S4-1, proximity analysis: analyze the proximity of the real-time data of device A to all preset data reference ranges, and find the data reference range Q with the largest proximity; S4-2, instruction selection and delivery: select a data reference range W with a risk index higher than the risk index corresponding to the data reference range Q, and the risk indexes corresponding to the data reference range W and the data reference range Q are adjacent, then retrieve the emergency instruction corresponding to the risk index of the data reference range W from the instruction library, and deliver it to device A. 2.The edge computing gateway of claim 1, wherein: When it is determined that the gateway hardware layer is faulty, a gateway warning is sent to the control center; when it is determined that the data transmission layer is faulty, a network warning is sent to the control center, and data transmission to the control center is suspended. 3.The edge computing gateway of claim 1, wherein: The system layer further comprises a pre-variation database, the pre-variation database stores a plurality of device pre-variation data ranges corresponding to emergency instructions one by one, in step S4-2, after the system layer delivers the emergency instruction to device A, the real-time data of device A is continuously received, when the real-time data of device A does not conform to the device pre-variation data range corresponding to the emergency instruction within a set time, it is determined that the device is out of control, the system layer sends an emergency warning to the control center, at the same time, the system layer retrieves the emergency instruction with the highest risk index from the instruction library and delivers it to device A.

4. The edge computing gateway of claim 1, wherein: The device fingerprint library further stores dynamic fingerprints, the dynamic fingerprints include historical behavior data of a plurality of devices, the historical behavior data includes data evolution trend, data abnormality period and data abnormality time.

5. The edge computing gateway of claim 4, wherein: In step S3, when the real-time data of device A is not within any corresponding data reference range, it is first determined whether the abnormal data of device A conforms to the historical behavior data, when it does not conform, the multi-dimensional fault diagnosis operation is performed again, when it conforms, no other operation is needed.

Images