Smart cloud box detection system based on edge computing

Abstract

The invention provides a smart cloud box detection system based on edge computing, and relates to the field of electric digital data processing, the smart cloud box detection system comprises a data anomaly detection module, an image acquisition module, an edge computing module and an alarm control module, the data anomaly detection module is used for carrying out anomaly detection on working data of a smart cloud box, and the image acquisition module is used for carrying out image acquisition; the image acquisition module is used for acquiring image information of a smart cloud box hardware facility, the edge calculation module is used for performing defect detection on the image information, and the alarm control module is used for performing alarm analysis on a data detection result and an equipment detection result and implementing a control instruction issued by a cloud server; according to the system, the image information of the hardware facility is acquired and analyzed, and the defect analysis is performed on the hardware equipment in combination with the operation data, so that the detection result can be directly given at the terminal, and the operation pressure of the cloud server is reduced.

Description

Technical Field

[0001] The present invention relates to the field of electrical digital data processing, and in particular to a smart cloud box detection system based on edge computing. Background Art

[0002] A smart cloud box usually refers to a device or solution that combines cloud computing and smart hardware. It uses the computing power of the cloud and the processing functions of local hardware to provide intelligent services and data management. Since the hardware facilities of the smart cloud box are exposed to the outdoors for a long time and are prone to damage, it is necessary to conduct timely inspections of the hardware to ensure the normal operation of the smart cloud box.

[0003] The foregoing discussion of the background art is intended only to facilitate an understanding of the present invention. This discussion does not acknowledge or admit that any of the material referred to is part of the common general knowledge.

[0004] Now many authorization systems have been developed. After a lot of searching and reference, it is found that the existing authorization systems are such as the system disclosed in the publication number CN117058460A. These system methods generally include: using enhanced shuffle network to improve the backbone network of the target detection model, using bolt data set to train the lightweight target detection model, and then converting and quantizing the format to obtain a lightweight model that adapts to the edge computing hardware platform; using the number of cores that match the hardware resource specifications in the edge computing hardware platform, and coordinating the lightweight model based on the real-time acquired bolt image to quickly obtain real-time corrosion detection results. However, this system only performs detection and analysis through images, and is only suitable for corrosion detection, not for smart cloud box detection. Summary of the invention

[0005] The purpose of the present invention is to propose a smart cloud box detection system based on edge computing to address the existing deficiencies.

[0006] The present invention adopts the following technical solution:

[0007] A smart cloud box detection system based on edge computing, including a data anomaly detection module, an image acquisition module, an edge computing module and an alarm control module;

[0008] The data anomaly detection module is used to perform anomaly detection on the working data of the smart cloud box, the image acquisition module is used to collect image information of the hardware facilities of the smart cloud box, the edge computing module is used to perform defect detection on the image information, and the alarm control module is used to perform alarm analysis on the data detection results and the equipment detection results and implement the control instructions issued by the cloud server;

[0009] The data anomaly detection module includes a conventional feature setting unit, an abnormal phenomenon capture unit and an abnormal type judgment unit. The conventional feature setting unit is used to set the data conventional features of all running items of the smart cloud box. The abnormal phenomenon capture unit is used to capture individual data that does not conform to the conventional features from the real-time running data. The abnormal feature judgment unit is used to analyze the individual data and judge the data anomaly type to which it belongs;

[0010] The image acquisition module includes a video shooting unit, an image interception unit and an information storage unit. The video shooting unit is used to shoot video information containing the hardware facilities of the smart cloud box. The image interception unit intercepts image information for analysis from the video information. The information storage unit is used to store the intercepted image information.

[0011] The edge computing module includes a device identification unit, a model analysis unit and a defect classification unit. The device identification unit is used to identify the device type in the image. The model analysis unit selects a corresponding defect analysis model based on the device type to perform defect analysis on the image. The defect classification unit performs defect classification on the data output by the model analysis unit.

[0012] The alarm control module includes a detection matching unit, an alarm feedback unit and a control execution unit, wherein the detection matching unit is used to match the data detection result with the image detection result, the alarm feedback unit issues an alarm based on the matching result, and the control execution unit is used to receive a control instruction and control the device;

[0013] Further, the model analysis unit includes a model selection processor, a model training processor and a defect analysis processor, the model selection processor selects a corresponding analysis model based on the identified hardware device type, the model training processor is used to perform data training on each analysis model to improve the analysis capability of the model, and the defect analysis processor is used to execute the selected analysis model to perform defect analysis on the frame image;

[0014] Further, the defect classification unit includes a defect data register and a defect mapping processor, the defect data register is used to store standard data information of all defects, and the defect mapping processor maps the data output by the analysis model to the corresponding defect type;

[0015] Further, when the defect data satisfies the following formula, the defect mapping processor maps the defect data to the corresponding defect type:

[0016] b imin ≤a i ≤ b imax , i=1,2,3,…,n;

[0017] Among them, ai represents the value of the i-th dimension in the defect data, [b imin , b imax ] is the i-th dimension interval of defect type standard data, and n is the dimension of defect data;

[0018] The defect mapping processor calculates the defect index P of the mapped defect type according to the following formula:

[0019]

[0020] Further, the detection and matching unit includes a type matching processor and a correction calculation processor, the type matching processor is used to match the abnormality type and the defect type, and the correction calculation processor calculates the alarm value based on the matching relationship;

[0021] The type matching processor sets a matching relationship between each defect type and anomaly type, each defect type has a matching relationship with multiple anomaly types, and each matching relationship is preset with a corresponding correction coefficient. The type matching processor selects an anomaly type with a maximum correction coefficient matching relationship from the detected anomaly types as the anomaly type that actually matches the defect type;

[0022] The correction calculation processor calculates the alarm value Q according to the following formula:

[0023]

[0024] Wherein, k is the correction coefficient between matching relationships.

[0025] The beneficial effects achieved by the present invention are:

[0026] This system identifies the hardware type in the image, uses the corresponding model to perform defect analysis to obtain the defect type, and combines the abnormal type of equipment operation data to obtain the alarm value. It can obtain more accurate detection results. The entire analysis process uses edge computing, which can effectively reduce the detection pressure of the cloud server.

[0027] To further understand the features and technical contents of the present invention, please refer to the following detailed description and drawings of the present invention. However, the drawings provided are only for reference and description and are not intended to limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 It is a schematic diagram of the overall structural framework of the present invention;

[0029] Figure 2 This is a schematic diagram of the data anomaly detection module of the present invention;

[0030] Figure 3This is a schematic diagram of the image acquisition module of the present invention;

[0031] Figure 4 This is a schematic diagram of the edge computing module of the present invention;

[0032] Figure 5 This is a schematic diagram of the alarm control module of the present invention. DETAILED DESCRIPTION

[0033] The following is an explanation of the embodiments of the present invention through specific embodiments. Those skilled in the art can understand the advantages and effects of the present invention from the contents disclosed in this specification. The present invention can be implemented or applied through other different specific embodiments, and the details in this specification can also be modified and changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. In addition, the drawings of the present invention are only simple schematic illustrations and are not depicted according to actual sizes. It is stated in advance. The following embodiments will further explain the relevant technical contents of the present invention in detail, but the disclosed contents are not intended to limit the scope of protection of the present invention.

[0034] Embodiment 1.

[0035] This embodiment provides a smart cloud box detection system based on edge computing, combined with Figure 1 , including data anomaly detection module, image acquisition module, edge computing module and alarm control module;

[0036] The data anomaly detection module is used to perform anomaly detection on the working data of the smart cloud box, the image acquisition module is used to collect image information of the hardware facilities of the smart cloud box, the edge computing module is used to perform defect detection on the image information, and the alarm control module is used to perform alarm analysis on the data detection results and the equipment detection results and implement the control instructions issued by the cloud server;

[0037] The data anomaly detection module includes a conventional feature setting unit, an abnormal phenomenon capture unit and an abnormal type judgment unit. The conventional feature setting unit is used to set the data conventional features of all running items of the smart cloud box. The abnormal phenomenon capture unit is used to capture individual data that does not conform to the conventional features from the real-time running data. The abnormal feature judgment unit is used to analyze the individual data and judge the data anomaly type to which it belongs;

[0038] The image acquisition module includes a video shooting unit, an image interception unit and an information storage unit. The video shooting unit is used to shoot video information containing the hardware facilities of the smart cloud box. The image interception unit intercepts image information for analysis from the video information. The information storage unit is used to store the intercepted image information.

[0039] The edge computing module includes a device identification unit, a model analysis unit and a defect classification unit. The device identification unit is used to identify the device type in the image. The model analysis unit selects a corresponding defect analysis model based on the device type to perform defect analysis on the image. The defect classification unit performs defect classification on the data output by the model analysis unit.

[0040] The alarm control module includes a detection matching unit, an alarm feedback unit and a control execution unit, wherein the detection matching unit is used to match the data detection result with the image detection result, the alarm feedback unit issues an alarm based on the matching result, and the control execution unit is used to receive a control instruction and control the device;

[0041] The model analysis unit includes a model selection processor, a model training processor and a defect analysis processor. The model selection processor selects a corresponding analysis model based on the identified hardware device type. The model training processor is used to perform data training on each analysis model to improve the analysis capability of the model. The defect analysis processor is used to execute the selected analysis model to perform defect analysis on the frame image.

[0042] The defect classification unit includes a defect data register and a defect mapping processor, wherein the defect data register is used to store standard data information of all defects, and the defect mapping processor maps the data output by the analysis model to the corresponding defect type;

[0043] When the defect data satisfies the following formula, the defect mapping processor maps the defect data to the corresponding defect type:

[0044] b imin ≤a i ≤ b imax , i=1,2,3,…,n;

[0045] Among them, a i represents the value of the i-th dimension in the defect data, [b imin , b imax ] is the i-th dimension interval of defect type standard data, and n is the dimension of defect data;

[0046] The defect mapping processor calculates the defect index P of the mapped defect type according to the following formula:

[0047]

[0048] The detection and matching unit includes a type matching processor and a correction calculation processor, wherein the type matching processor is used to match the abnormality type with the defect type, and the correction calculation processor calculates the alarm value based on the matching relationship;

[0049] The type matching processor sets a matching relationship between each defect type and anomaly type, each defect type has a matching relationship with multiple anomaly types, and each matching relationship is preset with a corresponding correction coefficient. The type matching processor selects an anomaly type with a maximum correction coefficient matching relationship from the detected anomaly types as the anomaly type that actually matches the defect type;

[0050] The correction calculation processor calculates the alarm value Q according to the following formula:

[0051]

[0052] Wherein, k is the correction coefficient between matching relationships.

[0053] Embodiment 2.

[0054] This embodiment includes all the contents of the first embodiment, and provides a smart cloud box detection system based on edge computing, including a data anomaly detection module, an image acquisition module, an edge computing module, and an alarm control module;

[0055] The data anomaly detection module is used to perform anomaly detection on the working data of the smart cloud box, the image acquisition module is used to collect image information of the hardware facilities of the smart cloud box, the edge computing module is used to perform defect detection on the image information, and the alarm control module is used to perform alarm analysis on the data detection results and the equipment detection results and implement the control instructions issued by the cloud server;

[0056] Combination Figure 2 The data anomaly detection module includes a conventional feature setting unit, an abnormal phenomenon capture unit and an abnormal type judgment unit. The conventional feature setting unit is used to set the data conventional features of all running projects of the smart cloud box. The abnormal phenomenon capture unit is used to capture individual data that does not conform to the conventional features from the real-time running data. The abnormal feature judgment unit is used to analyze the individual data and judge the data anomaly type to which it belongs;

[0057] Combination Figure 3 , the image acquisition module includes a video shooting unit, an image interception unit and an information storage unit, the video shooting unit is used to shoot video information containing the hardware facilities of the smart cloud box, the image interception unit intercepts image information for analysis from the video information, and the information storage unit is used to save the intercepted image information;

[0058] Combination Figure 4, the edge computing module includes a device identification unit, a model analysis unit and a defect classification unit, the device identification unit is used to identify the device type in the image, the model analysis unit selects a corresponding defect analysis model based on the device type to perform defect analysis on the image, and the defect classification unit performs defect classification on the data output by the model analysis unit;

[0059] Combination Figure 5 The alarm control module includes a detection and matching unit, an alarm feedback unit and a control execution unit. The detection and matching unit is used to match the data detection result and the image detection result. The alarm feedback unit issues an alarm based on the matching result. The control execution unit is used to receive a control instruction and control the device.

[0060] The conventional feature setting unit includes a setting interaction processor and a setting content register, wherein the setting interaction processor is used to provide a setting interface for editing and adding or reducing conventional feature information, and the setting content register is used to store the set conventional feature information;

[0061] The abnormal phenomenon capture unit includes a real-time comparison processor, an abnormal extraction processor and an abnormal packaging processor, wherein the real-time comparison processor is used to compare the real-time operation data with the conventional feature information, the abnormal extraction processor is used to extract the operation data that does not conform to the conventional feature information, and the abnormal packaging processor is used to package the flag information in the extracted operation data;

[0062] The mark information of the operation data includes the operation time, operation type and difference characteristics;

[0063] The abnormality judgment unit includes an individual filtering processor, an abnormality information register and an abnormality classification processor, wherein the individual filtering processor is used to filter occasional individual abnormal data packets and aggregate and send the individual abnormal data packets with consistency to the abnormality classification processor, the abnormality information register is used to store the judgment parameters of the abnormality type, and the abnormality classification processor analyzes the aggregated individual abnormal data packets based on the judgment parameters and outputs the abnormality type information;

[0064] The video shooting unit includes a patrol control processor, a network shooting processor and a video transmission processor, wherein the patrol control processor is used to control the rotation of the shooting lens, the network shooting processor is used to control the shooting lens to shoot video information, and the video transmission processor is used to transmit the shot video information to the image capture unit;

[0065] The image capture unit includes a video source buffer, an extraction control processor and a frame extraction processor, wherein the video source buffer is used to receive and temporarily store video information, the extraction control processor is used to send an extraction signal to the frame extraction processor, and the frame extraction processor is used to extract a frame image from the video information;

[0066] The information storage unit includes a frame image register and a frame image management processor, wherein the frame image register is used to store the extracted frame image, and the frame image management processor is used to control and manage the process of sending the frame image to the edge computing module;

[0067] The device identification unit includes a contour extraction processor and a device mapping processor, wherein the contour extraction processor is used to extract contour information in the frame image, and the device mapping processor maps to a corresponding hardware device based on edge information;

[0068] The model analysis unit includes a model selection processor, a model training processor and a defect analysis processor. The model selection processor selects a corresponding analysis model based on the identified hardware device type. The model training processor is used to perform data training on each analysis model to improve the analysis capability of the model. The defect analysis processor is used to execute the selected analysis model to perform defect analysis on the frame image.

[0069] The defect classification unit includes a defect data register and a defect mapping processor, wherein the defect data register is used to store standard data information of all defects, and the defect mapping processor maps the data output by the analysis model to the corresponding defect type;

[0070] The detection and matching unit includes a type matching processor and a correction calculation processor, wherein the type matching processor is used to match the abnormality type with the defect type, and the correction calculation processor calculates the alarm value based on the matching relationship;

[0071] The alarm feedback unit includes an alarm judgment processor and an alarm transmission processor, wherein the alarm judgment processor judges whether an alarm is required based on the alarm value, and the alarm transmission processor is used to send alarm information to the cloud server;

[0072] The control execution unit includes an instruction receiving processor and an instruction execution processor, wherein the instruction receiving processor is used to receive a control instruction sent by the cloud server, and the instruction execution processor controls the operation of the hardware device by executing the control instruction;

[0073] The training process of the model training processor includes the following steps:

[0074] S1. Initialize model parameters: use the pre-trained model to initialize parameters;

[0075] S2. Define the loss function: including classification loss and confidence loss;

[0076] S3, forward propagation: pass the input image through the model to obtain the predicted defect data;

[0077] S4. Calculate loss: Calculate loss based on prediction results and true annotations;

[0078] S5, back propagation and parameter update: calculate the gradient through the back propagation algorithm and use the optimizer to update the model parameters;

[0079] S6, iterative training: repeat the above steps until the loss function converges or reaches the preset number of training rounds;

[0080] The analysis process of the defect analysis processor includes the following steps:

[0081] S21, input image preprocessing: perform the same preprocessing on the input image as in training;

[0082] S22, forward propagation: input the preprocessed image into the model and continuously pass the processed data forward;

[0083] S23, output result: finally output defect data of the target object;

[0084] The defect data of the target object output by the analysis model is multi-dimensional data, and the data value range of each dimension is between 0 and 1. When the defect data satisfies the following formula, the defect mapping processor maps the defect data to the corresponding defect type:

[0085] b imin ≤a i ≤ b imax , i=1,2,3,…,n;

[0086] Among them, a i represents the value of the i-th dimension in the defect data, [b imin , b imax ] is the i-th dimension interval of defect type standard data, and n is the dimension of defect data;

[0087] The defect mapping processor calculates the defect index P of the mapped defect type according to the following formula:

[0088]

[0089] The type matching processor sets a matching relationship between each defect type and anomaly type, each defect type has a matching relationship with multiple anomaly types, and each matching relationship is preset with a corresponding correction coefficient. The type matching processor selects an anomaly type with a maximum correction coefficient matching relationship from the detected anomaly types as the anomaly type that actually matches the defect type;

[0090] The correction calculation processor calculates the alarm value Q according to the following formula:

[0091]

[0092] Wherein, k is the correction coefficient between matching relationships;

[0093] When the alarm value is greater than the threshold, the alarm judgment processor determines that alarm processing is required;

[0094] The i appearing in the above text is an ordinal number used to indicate a sequence number.

[0095] The contents disclosed above are only preferred feasible embodiments of the present invention, and do not limit the protection scope of the present invention. Therefore, all equivalent technical changes made using the contents of the present invention specification and drawings are included in the protection scope of the present invention. In addition, the elements therein can be updated as technology develops.

Claims

1. A smart cloud box detection system based on edge computing, characterized in that: It includes data anomaly detection module, image acquisition module, edge computing module and alarm control module; The data anomaly detection module is used to perform anomaly detection on the working data of the smart cloud box, the image acquisition module is used to collect image information of the hardware facilities of the smart cloud box, the edge computing module is used to perform defect detection on the image information, and the alarm control module is used to perform alarm analysis on the data detection results and the equipment detection results and implement the control instructions issued by the cloud server; The data anomaly detection module includes a conventional feature setting unit, an abnormal phenomenon capture unit and an abnormal type judgment unit. The conventional feature setting unit is used to set the data conventional features of all running items of the smart cloud box. The abnormal phenomenon capture unit is used to capture individual data that does not conform to the conventional features from the real-time running data. The abnormal feature judgment unit is used to analyze the individual data and judge the data anomaly type to which it belongs; The image acquisition module includes a video shooting unit, an image interception unit and an information storage unit. The video shooting unit is used to shoot video information containing the hardware facilities of the smart cloud box. The image interception unit intercepts image information for analysis from the video information. The information storage unit is used to store the intercepted image information. The edge computing module includes a device identification unit, a model analysis unit and a defect classification unit. The device identification unit is used to identify the device type in the image. The model analysis unit selects a corresponding defect analysis model based on the device type to perform defect analysis on the image. The defect classification unit performs defect classification on the data output by the model analysis unit. The alarm control module includes a detection matching unit, an alarm feedback unit and a control execution unit. The detection matching unit is used to match the data detection results and the image detection results. The alarm feedback unit issues an alarm based on the matching results. The control execution unit is used to receive control instructions and control the device.

2. The smart cloud box detection system based on edge computing as claimed in claim 1, characterized in that: The model analysis unit includes a model selection processor, a model training processor and a defect analysis processor. The model selection processor selects a corresponding analysis model based on the identified hardware device type. The model training processor is used to perform data training on each analysis model to improve the analysis capability of the model. The defect analysis processor is used to execute the selected analysis model to perform defect analysis on the frame image.

3. The smart cloud box detection system based on edge computing as claimed in claim 2, characterized in that: The defect classification unit includes a defect data register and a defect mapping processor. The defect data register is used to store standard data information of all defects. The defect mapping processor maps the data output by the analysis model to the corresponding defect type.

4. The smart cloud box detection system based on edge computing as claimed in claim 3, characterized in that: When the defect data satisfies the following formula, the defect mapping processor maps the defect data to the corresponding defect type: b imin ≤a i ≤b imax ,i=1,2,3,…,n; Among them, a i represents the value of the i-th dimension in the defect data, [b imin , b imax ] is the i-th dimension interval of defect type standard data, and n is the dimension of defect data; The defect mapping processor calculates the defect index P of the mapped defect type according to the following formula:

5. The edge computing-based smart cloud box detection system according to claim 4, characterized in that: The detection and matching unit includes a type matching processor and a correction calculation processor, wherein the type matching processor is used to match the abnormality type with the defect type, and the correction calculation processor calculates the alarm value based on the matching relationship; The type matching processor sets a matching relationship between each defect type and anomaly type, each defect type has a matching relationship with multiple anomaly types, and each matching relationship is preset with a corresponding correction coefficient. The type matching processor selects an anomaly type with a maximum correction coefficient matching relationship from the detected anomaly types as the anomaly type that actually matches the defect type; The correction calculation processor calculates the alarm value Q according to the following formula: Wherein, k is the correction coefficient between matching relationships.

Images