CN116796256A - An Internet of Things data analysis and anomaly detection method and system - Google Patents
An Internet of Things data analysis and anomaly detection method and system Download PDFInfo
- Publication number
- CN116796256A CN116796256A CN202211702367.9A CN202211702367A CN116796256A CN 116796256 A CN116796256 A CN 116796256A CN 202211702367 A CN202211702367 A CN 202211702367A CN 116796256 A CN116796256 A CN 116796256A
- Authority
- CN
- China
- Prior art keywords
- data
- internet
- clustering
- things
- sensor
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
- G16Y40/00—IoT characterised by the purpose of the information processing
- G16Y40/10—Detection; Monitoring
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
- G16Y40/00—IoT characterised by the purpose of the information processing
- G16Y40/20—Analytics; Diagnosis
Landscapes
- Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Testing Or Calibration Of Command Recording Devices (AREA)
Abstract
The invention discloses a method and a system for analyzing data and detecting abnormality of the Internet of things, and belongs to the technical field of the Internet of things. According to the method, firstly, sensor data of the Internet of things are clustered on a time axis, a machine learning model is built based on clusters generated by clustering, actual observation data are compared with data predicted by the machine learning model to judge whether abnormality exists in a network, and the machine learning model adopts a self-information entropy/mutual-information entropy model, a CNN+transducer model and an LSTM+transducer model. The data analysis and anomaly detection system mainly comprises a data clustering module, a data dimension reduction module, a machine learning module and an anomaly detection module, and the anomaly detection method is achieved. The method can analyze massive data of the Internet of things, and simultaneously can rapidly detect abnormality aiming at different sensor time series data without marking, thereby effectively solving the problem of data explosion; compared with the traditional method, the method can also automatically monitor the abnormal condition of the change of the network topology structure.
Description
Technical Field
The invention relates to a method and a system for analyzing data and detecting abnormality of the Internet of things, and belongs to the technical field of the Internet of things.
Background
At present, the technology of the internet of things is widely applied in the fields of intelligent power grids, intelligent environmental protection, intelligent water affairs, intelligent traffic and the like, in the running process of the internet of things, each sensor can generate data far larger than that of a traditional service system, the mass data can bring about small difficulty to the abnormal judgment in the internet of things, a traditional centralized service platform mostly adopts a simple threshold method to judge the abnormality, the method cannot fully utilize a large amount of internet of things service data and potential relations among the data, the detection precision of the abnormality is lower, and because the data quantity is large, effective model description is difficult to build on the data, and the difficulty is further brought to the judgment of the abnormal state of the internet of things.
In addition, when the topological structure of the internet of things system changes, such as power system load changes, water service system pipelines change and the like, the state of the internet of things cannot be updated in time, and when relevant corresponding data changes, larger errors can be generated by utilizing the traditional fixed threshold judgment. Therefore, a new method is needed to judge the data generated by a large number of sensors in the internet of things and determine the occurrence of abnormal situations.
Disclosure of Invention
Technical problem to be solved by the invention
The invention provides a method and a system for analyzing and detecting data of the Internet of things, aiming at the problems of low utilization efficiency and no labeling of the data of the Internet of things.
Technical proposal
In order to achieve the above purpose, the technical scheme provided by the invention is as follows:
a data analysis and anomaly detection method of the Internet of things comprises the following steps:
step 1, clustering sensor data on a time axis, and when the clustering time exceeds a certain value, firstly performing dimension reduction on the data and then clustering;
and 2, establishing a machine learning model based on the clusters generated by the clustering in the step 1, and comparing actual observation data with data predicted by the machine learning model to judge whether an abnormality exists in the network.
Further, the method for judging the abnormality in the step 2 is that
Step 21, regarding the data belonging to one sensor in the cluster as a group, and solving the mean point of the data for all the data in the group;
step 22, equally dividing the interval between the mean point and the data point with the largest distance in the group of data into N areas, sequentially assigning each area as 1,2, … and N, equally dividing the interval between the mean point and the data point with the smallest distance in the group of data into M areas, sequentially assigning each area as-1, -2, … and M;
step 23, counting the number of data points falling into each interval in the group of data, calculating the occurrence probability of the data in each interval, and marking as P i ，i＝-M,…,-1,0,1,…N
The average self-information entropy of the set of data is recorded as
Step 24, obtaining the average self-information entropy of the group of data, clustering the data at the moment k, reducing the dimension, and obtaining the average self-information entropy of a certain group of data by using the same method when the actual sensor network is in operation
When H is k And when the absolute value of the difference between (X) and H (X) exceeds a preset threshold alpha, the sensor network at the moment k is considered to be abnormal.
Further, the method for judging the abnormality in the step 2 is as follows:
step 21, data X and Y belonging to two sensors in the cluster are respectively calculated to obtain mean value points of the two groups of data;
step 22, equally dividing the interval between the mean point and the data point with the largest distance in the group of data into N areas, sequentially assigning each area as 1,2, … and N, equally dividing the interval between the mean point and the data point with the smallest distance in the group of data into M areas, sequentially assigning each area as-1, -2, … and M;
step 23, counting the number of data points falling into each interval in the group of data, calculating the occurrence probability of the data in each interval, and marking as Q i ，i＝-M,…,-1,0,1,…N
The mutual information entropy of the two groups of data X and Y is recorded as
Wherein Q is ij Representing joint probabilities of data occurrence in the i, j-th interval;
step 24, obtaining mutual information entropy of two groups of data, clustering the data at k moment and reducing dimension when the actual sensor network operates, and obtaining average self-information entropy of a certain group of data by using the same method
When H is k And when the absolute value of the difference between (X, Y) and H (X, Y) exceeds a preset threshold beta, the sensor corresponding to the sensor network at the moment k is considered to be abnormal.
Further, the method for judging the abnormality in the step 2 is as follows:
step 21, establishing a CNN+transducer model for cluster data generated by a plurality of clusters;
step 22, when the absolute value of the difference between the actual observed data and the model predicted data obtained by training in step 21 is higher than a certain predetermined threshold, the sensor network is considered to be abnormal.
Further, the method for judging the abnormality in the step 2 is as follows:
step 21, establishing an LSTM+transducer model for cluster data generated by a plurality of clusters;
step 22, when the absolute value of the difference between the actual observed data and the model predicted data obtained by training in step 21 is higher than a certain predetermined threshold, the sensor network is considered to be abnormal.
The invention relates to an Internet of things data analysis and anomaly detection system for realizing an Internet of things data analysis and anomaly detection method, which mainly comprises the following steps: the system comprises a data clustering module, a data dimension reduction module, a machine learning module and an abnormality detection module, wherein:
the data clustering module realizes clustering of sensor data in the abnormality detection method
The data analysis dimension reduction module is used for realizing analysis of sensor data and dimension reduction of high-dimension sensor data
The machine learning module establishes a calculation learning model of the sensor data through a machine learning method
The anomaly detection module uses a machine learning model to determine whether anomalies exist in the sensor network.
Advantageous effects
The method can analyze mass internet of things data, and effectively solves the problem of data explosion;
the method can find abnormal data aiming at different sensor time series data without labeling, and can rapidly detect the abnormality of the Internet of things.
Compared with the traditional method, the method can automatically monitor the abnormal condition of the change of the network topology structure.
Drawings
FIG. 1 is a step diagram of a method for analyzing data and detecting anomalies of the Internet of things according to the present invention;
FIG. 2 is a diagram of a CNN+ transducer model constructed in accordance with the present invention;
FIG. 3 is a diagram of an LSTM+transducer model constructed in accordance with the present invention;
fig. 4 is a block diagram of an internet of things data analysis and anomaly detection system according to the present invention.
Detailed Description
For a further understanding of the present invention, reference should be made to the following detailed description of the invention, taken in conjunction with the accompanying drawings and detailed description.
As shown in fig. 1, the data analysis and anomaly detection method of the internet of things comprises the following steps:
step 1, clustering sensor data on a time axis, wherein the clustering purpose is to separate time sequence data in time by adopting a k-means method and a KNN method, data in each cluster obtained by each cluster represents a stable topological structure, the change of the topological structure of the Internet of things can be responded quickly through the clustering, and when the data of the Internet of things has no clustering result in a certain time, the data is subjected to dimension reduction at first and then is clustered. The dimension reduction method can adopt a linear dimension reduction method and a nonlinear dimension reduction method, in the embodiment, the linear dimension reduction method adopts PCA, and the nonlinear dimension reduction method adopts t-SNE.
And 2, establishing a related data model based on the clusters generated by the clustering in the step 1 to judge network abnormality. The invention mainly adopts two types of methods:
the first judgment method comprises the following steps: in the clusters generated by each cluster, data belonging to one sensor is regarded as a group, and self-information entropy of the group of data or mutual information entropy of multiple groups of data is calculated whether a preset threshold value is exceeded by a corresponding mean value. If the threshold is exceeded, the method is abnormal, and the method for calculating the self-information entropy of a certain group of data comprises the following steps:
step 1, calculating the mean point of the data for all the data in the group.
Step 2, equally dividing the interval between the mean value point and the data point with the largest distance into N areas, and sequentially assigning 1,2, … and N to each area; the interval between the mean point and the data point with the smallest distance is divided into M areas, and each area is assigned with-1, -2, … and M in sequence.
Step 3, counting the number of data points falling into each interval in the group of data, calculating the occurrence probability of the data in each interval, and marking as P i ，i＝-M,…,-1,0,1,…N
The average self-information entropy of the set of data is recorded as
Wherein P is i The probability of occurrence of the data points in the ith interval is obtained by calculating the ratio of the number of the data points in the region to the total number of the data points.
Step 4, after obtaining the average self-information entropy of the group of data, when the actual sensor network operates, obtaining the average self-information entropy of a certain group of data by using the same method after data clustering and dimension reduction at time k
When H is k When the absolute value of the difference between (X) and H (X) exceeds a predetermined threshold α, the sensor network at time k is considered to be abnormal, and in this embodiment, the logarithm preference is based on 2, and the threshold α is 0.5.
The data average value and the self-information entropy in the group need to be recalculated after a certain time t so as to achieve the effect of real-time updating.
The method for calculating mutual information entropy between the groups of data comprises the following steps:
and step 1, calculating the mean value point of each group of data.
Step 2, equally dividing the interval between the mean value point and the data point with the largest distance in the group of data into N areas, and sequentially assigning each area as 1,2, … and N; the interval between the mean point and the data point with the smallest distance in the group of data is divided into M areas, and each area is assigned with-1, -2, … and M in sequence.
Step 3, counting the number of data points falling into each interval, calculating the data occurrence probability of each interval in the group of data, and marking as P i ，i＝-M,…,-1,0,1,…N；
Counting the number of data points falling into each interval in the group of data, calculating the occurrence probability of the data in each interval, and marking as P i ，i＝-M,…,-1,0,1,…N；
The mutual information entropy of the two groups of data X and Y is recorded as
Wherein Q is ij Indicating the joint probability of data occurrence in the i, j-th interval.
Step 4, at a certain time k, if the mutual information entropy of the two kinds of sensor data is
When H k When the absolute value of the difference between (X, Y) and H (X, Y) exceeds a predetermined threshold β, it is indicated that the sensor data corresponding to the moment is abnormal, and in this embodiment, the logarithm preference is based on 2, and the threshold β is 0.5.
And meanwhile, the self information entropy, the intra-group data average value and the mutual information entropy need to be recalculated after a certain time t so as to achieve the effect of real-time updating.
And a judging method II: and building machine learning model training by using the sensor raw data. The present embodiment uses a transducer model to train the raw data. Specifically, the combination of CNN (bottom layer) +transducer or LSTM (bottom layer) +transducer can be used
As shown in fig. 2, the CNN + transducer model includes four layers, the lowest layer being the input layer and above in turn the attention layer, likelihood layer and target layer,
in the figure, input Z t (t=1, 2, 3) is data of three time points generated in the clustering step, corresponding to data of a certain time period of the sensor network, configuring parameters related to a CNN model and a transducer model, and in the embodiment, obtaining a final predicted value through linear transformation twice by adopting a 3*3 convolution kernel on the CNN layer; the number of layers of the encoder/decoder of the transducer model is 6.
Will input data Z t Sequentially inputting a CNN model and a transducer model for calculation to obtain attention layer data o t (t＝1,2,3)：o t ＝Transformer(CNN(Z t-1 ,Z t ,…))
Assuming a target value z in each time step t Obeys the probability distribution l (z) t |θ t ) Probability distribution parameter θ by means of attention layer data t Calculation is performed, in the present embodiment, the probability distribution function l selects a gaussian function, θ t ～N(μ,σ 2 ) Mu and sigma 2 For the attention layer data o t Mean and variance of (c).
Under the condition that the observation data of the target layer are known, optimizing the network in the graph by maximizing the target function, and finally taking the network parameter which maximizes the target function to complete model training. In the present embodiment, the objective function is the sum Σ of log likelihood functions of the respective observations t l(z t |θ t )
As shown in fig. 3, the LSTM + transducer model comprises five layers, a bottom layer being an LSTM layer, an attention layer, a likelihood layer and a target layer in order above an input layer,
in the figure, input Z t (t=1, 2, 3) is the data of three time points generated in the previous step, corresponding to the data of a certain time period of the sensor network, configuring parameters related to the LSTM model and the transducer model, and in this embodiment, using tanh for the LSTM layer activation function; the number of layers of the encoder/decoder of the transducer model is 6.
Calculating the hidden state of the current time step by using an LSTM unit: h is a t ＝LSTM(Z t-1 ,h t-1 ),(t＝1,2,3)
The LSTM layer data is input into a transducer model, and the attention layer data o is calculated t (t＝1,2,3)：o t ＝Transformer(h t )
Assuming a target value z in each time step t Obeys the probability distribution l (z) t |θ t ) Probability distribution parameter θ by means of attention layer data t Calculation is performed, in the present embodiment, the probability distribution function l selects a gaussian function, θ t ～N(μ,σ 2 ) Mu and sigma 2 For the attention layer data o t Mean and variance of (c).
Optimizing the network in the graph by maximizing the objective function given the observation data of the target layerAnd finally, taking network parameters which maximize the objective function, and completing model training. In the present embodiment, the objective function is the sum Σ of log likelihood functions of the respective observations t l(z t |θ t )
The CNN+transducer model or LSTM+transducer model obtained by training the steps is used for predicting the actual data, and the actual observed data is set asPredicting data as z from a model t ' when the absolute value of the difference between the two is higher than a certain preset threshold, the sensor network is considered to be abnormal, in the embodiment, the difference is +.>Greater than |z t 5% of'.
As shown in fig. 4, an internet of things data analysis and anomaly detection system for implementing the internet of things data analysis and anomaly detection method of the present invention mainly includes: the system comprises a data clustering module, a data dimension reduction module, a machine learning module and an abnormality detection module.
The data clustering module is used for clustering the sensor data in the anomaly detection method; the data analysis dimension reduction module is used for realizing analysis of the sensor data and dimension reduction of the high-dimension sensor data; the machine learning module establishes a calculation learning model of the sensor data through a machine learning method; the anomaly detection module uses a machine learning model to determine whether anomalies exist in the sensor network.
The invention and its embodiments have been described above by way of illustration and not limitation, and the invention is illustrated in the accompanying drawings and described in the drawings in which the actual structure is not limited thereto. Therefore, if one of ordinary skill in the art is informed by this disclosure, the structural mode and the embodiments similar to the technical scheme are not creatively designed without departing from the gist of the present invention.
Claims (6)
1. A data analysis and anomaly detection method of the Internet of things comprises the following steps:
step S1, clustering sensor data on a time axis, and when the clustering time exceeds a certain value, firstly performing dimension reduction on the data and then clustering;
step S2, a machine learning model is built based on clusters generated by clustering in the step S1, and actual observation data is compared with data predicted by the machine learning model to judge whether an abnormality exists in the network.
2. The method for analyzing and detecting the anomaly of the internet of things according to claim 1, wherein the method for judging the anomaly of the network in the step S2 is as follows:
step S21, regarding the data belonging to one sensor in the cluster as a group, and solving the mean point of the data for all the data in the group;
step S22, equally dividing the interval between the mean value point and the data point with the largest distance in the group of data into N areas, sequentially assigning each area as 1,2, … and N, equally dividing the interval between the mean value point and the data point with the smallest distance in the group of data into M areas, sequentially assigning each area as-1, -2, … and M;
step S23, counting the number of data points falling into each interval in the group of data, calculating the occurrence probability of the data in each interval, and marking as P i ，i＝-M,…,-1,0,1,…N
Calculating the average self-information entropy of the set of data:
step S24, after obtaining the average self-information entropy of the group of data, when the actual sensor network operates, obtaining the average self-information entropy of a certain group of data by using the same method after data clustering and dimension reduction at time k
When H is k And when the absolute value of the difference between (X) and H (X) exceeds a preset threshold alpha, the sensor network at the moment k is considered to be abnormal.
3. The method for analyzing and detecting the anomaly of the internet of things according to claim 1, wherein the method for judging the anomaly of the network in the step S2 is as follows:
step S21, data X and Y belonging to two sensors in a cluster are respectively calculated to obtain mean value points of the two groups of data;
step S22, equally dividing the interval between the mean value point and the data point with the largest distance in the group of data into N areas, sequentially assigning each area as 1,2, … and N, equally dividing the interval between the mean value point and the data point with the smallest distance in the group of data into M areas, sequentially assigning each area as-1, -2, … and M;
step S23, counting the number of data points falling into each interval in the group of data, calculating the occurrence probability of the data in each interval, and marking as Q i ，i＝-M,…,-1,0,1,…N
Calculating mutual information entropy of the two groups of data X and Y:
wherein Q is ij Representing joint probabilities of data occurrence in the i, j-th interval;
step S24, obtaining mutual information entropy of the two groups of data, clustering the data at the moment k and reducing the dimension when the actual sensor network operates, and obtaining average self-information entropy of the two groups of data by using the same method
When H is k And when the absolute value of the difference between (X, Y) and H (X, Y) exceeds a preset threshold beta, the sensor corresponding to the sensor network at the moment k is considered to be abnormal.
4. The method for analyzing and detecting the anomaly of the internet of things according to claim 1, wherein the method for judging the anomaly of the network in the step S2 is as follows:
s21, establishing a CNN+transducer model for cluster data generated by a plurality of clusters;
step S22, when the absolute value of the difference between the actual observed data and the model predicted data obtained through training in step S21 is higher than a certain preset threshold, the sensor network is considered to be abnormal.
5. The method for analyzing and detecting the anomaly of the internet of things according to claim 1, wherein the method for judging the anomaly of the network in the step S2 is as follows:
s21, establishing an LSTM+transducer model for cluster data generated by a plurality of clusters;
step S22, when the absolute value of the difference between the actual observed data and the model predicted data obtained through training in step S21 is higher than a certain preset threshold, the sensor network is considered to be abnormal.
6. An internet of things data analysis and anomaly detection system for implementing the internet of things data analysis and anomaly detection method as claimed in any one of claims 1-5, wherein the system comprises a data clustering module, a data dimension reduction module, a machine learning module, and an anomaly detection module, wherein:
the data clustering module is used for clustering the sensor data in the anomaly detection method;
the data analysis dimension reduction module is used for realizing analysis of the sensor data and dimension reduction of the high-dimension sensor data;
the machine learning module establishes a calculation learning model of the sensor data through a machine learning method;
the anomaly detection module uses a machine learning model to determine whether anomalies exist in the sensor network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211702367.9A CN116796256A (en) | 2022-12-29 | 2022-12-29 | An Internet of Things data analysis and anomaly detection method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211702367.9A CN116796256A (en) | 2022-12-29 | 2022-12-29 | An Internet of Things data analysis and anomaly detection method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116796256A true CN116796256A (en) | 2023-09-22 |
Family
ID=88038141
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211702367.9A Pending CN116796256A (en) | 2022-12-29 | 2022-12-29 | An Internet of Things data analysis and anomaly detection method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116796256A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118690299A (en) * | 2024-06-19 | 2024-09-24 | 北京理工大学 | An anomaly detection method for spacecraft telemetry data |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160342903A1 (en) * | 2015-05-21 | 2016-11-24 | Software Ag Usa, Inc. | Systems and/or methods for dynamic anomaly detection in machine sensor data |
| CN112288034A (en) * | 2020-11-19 | 2021-01-29 | 江南大学 | Semi-supervised online anomaly detection method for wireless sensor network |
| CN113382413A (en) * | 2021-06-07 | 2021-09-10 | 西安电子科技大学 | Intelligent unmanned equipment networking anomaly detection method and system based on machine learning |
| CN115512152A (en) * | 2022-09-08 | 2022-12-23 | 武汉大学 | A ship trajectory classification method and system based on a combination of CNN and LSTM neural network |
-
2022
- 2022-12-29 CN CN202211702367.9A patent/CN116796256A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160342903A1 (en) * | 2015-05-21 | 2016-11-24 | Software Ag Usa, Inc. | Systems and/or methods for dynamic anomaly detection in machine sensor data |
| CN112288034A (en) * | 2020-11-19 | 2021-01-29 | 江南大学 | Semi-supervised online anomaly detection method for wireless sensor network |
| CN113382413A (en) * | 2021-06-07 | 2021-09-10 | 西安电子科技大学 | Intelligent unmanned equipment networking anomaly detection method and system based on machine learning |
| CN115512152A (en) * | 2022-09-08 | 2022-12-23 | 武汉大学 | A ship trajectory classification method and system based on a combination of CNN and LSTM neural network |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118690299A (en) * | 2024-06-19 | 2024-09-24 | 北京理工大学 | An anomaly detection method for spacecraft telemetry data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN119443922B (en) | An efficient and accurate pollution source monitoring quality control system and method | |
| CN110555551B (en) | Air quality big data management method and system for smart city | |
| CN117354171B (en) | Platform health condition early warning method and system based on Internet of things platform | |
| CN117640342B (en) | A method, device, equipment and medium for detecting abnormality of a power monitoring system | |
| CN111191855B (en) | Water quality abnormal event identification and early warning method based on pipe network multi-element water quality time sequence data | |
| CN111008662B (en) | Online monitoring data anomaly analysis method for power transmission line | |
| CN117744495B (en) | A multi-model driven life prediction method for extra-large bearings at different degradation stages | |
| CN114154567A (en) | Wind power plant station operation data anomaly identification method based on machine learning | |
| CN119416074A (en) | A machine learning water quality monitoring model based on random forest algorithm | |
| CN116244199A (en) | An abnormal detection method for operation and maintenance data based on multiple neural networks | |
| CN119612305A (en) | Intelligent analysis and early warning method and system for elevator emergency rescue data | |
| CN120185193A (en) | Intelligent power distribution cabinet state dynamic monitoring method and system | |
| CN119129772A (en) | A river section water quality prediction method based on combined machine learning model | |
| CN116796256A (en) | An Internet of Things data analysis and anomaly detection method and system | |
| CN117923331A (en) | Load control system and method based on crane hoisting | |
| CN112816211A (en) | Belt conveyor fault diagnosis method based on PSO-BP algorithm | |
| CN117688504B (en) | Internet of things abnormality detection method and device based on graph structure learning | |
| CN119830177A (en) | Intelligent equipment integrated management system for power grid distribution network inspection | |
| CN118038230A (en) | Distribution network defect identification system and method based on multi-algorithm fusion | |
| CN117306608A (en) | Foundation pit big data collection and intelligent monitoring method and its Internet of Things system | |
| CN110794799A (en) | Big data system with fault diagnosis function applied to industrial production | |
| CN117520900A (en) | Multi-factor precipitation level prediction method based on DQN_Multi-BiGRU and multi-head ProbSparse self-attention | |
| CN119167077A (en) | A fast fault location method for hot rolling mill based on FCM | |
| CN117709806B (en) | Cooperative multi-equipment abnormality automatic detection method and detection system | |
| CN119891947B (en) | Automatic monitoring method, device and equipment for photovoltaic power station and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| DD01 | Delivery of document by public notice |
Addressee: Sumo Intelligent Technology (Nanjing) Co.,Ltd. Document name: Notice of Deemed Failure to Entrust Patent Agency Addressee: Sumo Intelligent Technology (Nanjing) Co.,Ltd. Document name: Correction notice |
|
| DD01 | Delivery of document by public notice | ||
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |