CN113658415B - Early warning method and system of intelligent gateway - Google Patents
Early warning method and system of intelligent gateway Download PDFInfo
- Publication number
- CN113658415B CN113658415B CN202110875021.8A CN202110875021A CN113658415B CN 113658415 B CN113658415 B CN 113658415B CN 202110875021 A CN202110875021 A CN 202110875021A CN 113658415 B CN113658415 B CN 113658415B
- Authority
- CN
- China
- Prior art keywords
- fuzzy
- early warning
- value
- time sequence
- fuzzy subset
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 238000012549 training Methods 0.000 claims abstract description 106
- 238000011156 evaluation Methods 0.000 claims abstract description 95
- 238000012544 monitoring process Methods 0.000 claims abstract description 95
- 238000013528 artificial neural network Methods 0.000 claims abstract description 14
- 230000004044 response Effects 0.000 claims abstract description 7
- 238000012545 processing Methods 0.000 claims description 29
- 238000005457 optimization Methods 0.000 claims description 19
- 238000010606 normalization Methods 0.000 claims description 12
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 9
- 238000002372 labelling Methods 0.000 claims description 3
- 230000006870 function Effects 0.000 description 36
- 238000004891 communication Methods 0.000 description 32
- 230000005540 biological transmission Effects 0.000 description 23
- QVFWZNCVPCJQOP-UHFFFAOYSA-N chloralodol Chemical compound CC(O)(C)CC(C)OC(O)C(Cl)(Cl)Cl QVFWZNCVPCJQOP-UHFFFAOYSA-N 0.000 description 14
- 238000010586 diagram Methods 0.000 description 10
- 238000004519 manufacturing process Methods 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 4
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 238000003860 storage Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 229910000831 Steel Inorganic materials 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 230000035945 sensitivity Effects 0.000 description 2
- 239000010959 steel Substances 0.000 description 2
- 206010063385 Intellectualisation Diseases 0.000 description 1
- 241001131927 Placea Species 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000005059 dormancy Effects 0.000 description 1
- 238000004070 electrodeposition Methods 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 230000017525 heat dissipation Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 230000007958 sleep Effects 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/18—Status alarms
- G08B21/185—Electrical failure alarms
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/043—Architecture, e.g. interconnection topology based on fuzzy logic, fuzzy membership or fuzzy inference, e.g. adaptive neuro-fuzzy inference systems [ANFIS]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B31/00—Predictive alarm systems characterised by extrapolation or other computation using updated historic data
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L12/00—Data switching networks
- H04L12/66—Arrangements for connecting between networks having differing types of switching systems, e.g. gateways
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Emergency Management (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Automation & Control Theory (AREA)
- Pure & Applied Mathematics (AREA)
- Signal Processing (AREA)
- Computational Mathematics (AREA)
- Fuzzy Systems (AREA)
- Mathematical Analysis (AREA)
- Mathematical Optimization (AREA)
- Computer Networks & Wireless Communication (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Testing And Monitoring For Control Systems (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The application provides an intelligent gateway early warning method and system, wherein an intelligent early warning model based on a fuzzy neural network algorithm is deployed in the intelligent gateway, and the early warning method comprises the following steps: based on the intelligent early warning model after training, obtaining a state evaluation predicted value of the monitoring object at the next moment according to the fuzzy subset, a preset fuzzy rule corresponding to the fuzzy subset and the acquired input data set of the monitoring object; the fuzzy subset is obtained by dividing a training set acquired in advance through a clustering method; the training set is used for training the intelligent early warning model, the input data set comprises time sequence data of the current moment and a plurality of previous moments, and each time sequence data comprises a time sequence number value of a monitoring object and a state evaluation value corresponding to the time sequence number value; and comparing the state evaluation predicted value with a preset state threshold value, and early warning the running state of the monitored object at the next moment in response to the state evaluation predicted value being larger than the preset state threshold value.
Description
Technical Field
The application relates to the technical field of Internet of things, in particular to an early warning method and system of an intelligent gateway.
Background
Intelligent manufacturing has become an important direction of future manufacturing industry development, and the integration of new generation information technology and manufacturing industry is pushed to the full power, and manufacturing enterprises face transformation upgrading pressures in aspects of digitization, intellectualization and the like in the production process.
In the generation process, how to effectively monitor the running state of the equipment and discover equipment faults in time is also a key point for guaranteeing the efficient running of the production process. At present, a threshold value alarming mode is widely adopted in industrial production, the fault condition of equipment can only be passively reflected after the fault occurs, and the fault can not be timely avoided.
Accordingly, there is a need to provide an improved solution to the above-mentioned deficiencies of the prior art.
Disclosure of Invention
The purpose of the application is to provide an early warning method and system of an intelligent gateway, so as to solve or alleviate the problems existing in the prior art.
In order to achieve the above object, the present application provides the following technical solutions:
the application provides an intelligent gateway early warning method, wherein an intelligent early warning model based on a fuzzy neural network algorithm is deployed in the intelligent gateway, and the early warning method comprises the following steps: step S101, based on the intelligent early warning model after training, obtaining a state evaluation predicted value of the monitoring object at the next moment according to a fuzzy subset, a preset fuzzy rule corresponding to the fuzzy subset and an acquired input data set of the monitoring object; the state evaluation predicted value is used for representing the running state of the monitoring object at the next moment; the fuzzy subset is obtained by dividing a training set obtained in advance through a clustering method; the training set is used for training the intelligent early warning model, the input data set comprises time sequence data of the current moment and a plurality of previous moments, and each time sequence data comprises a time sequence value of the monitoring object and a state evaluation value corresponding to the time sequence value; and step S102, comparing the state evaluation predicted value with a preset state threshold value, and early warning the running state of the monitored object at the next moment in response to the state evaluation predicted value being larger than the preset state threshold value.
Preferably, step S101 includes: step S111, calculating a membership function value of each time sequence data and the corresponding fuzzy subset according to the membership function model of the fuzzy subset corresponding to the acquired input data set of the monitoring object; step S121, calculating the fitness of each fuzzy subset and the corresponding fuzzy rule according to the membership function value of each time sequence data and the corresponding fuzzy subset; step S131, carrying out normalization processing on the fitness of each fuzzy subset to obtain the weight of each fuzzy subset; step S141, according to the state prediction result of each fuzzy subset and the weight of each fuzzy subset, according to the formula:
obtaining a state evaluation predicted value of the monitoring object at the next moment; wherein, according to the independent prediction model:
obtaining a state prediction result of each fuzzy subset on the next moment of the monitoring object; wherein y is k+1 A state evaluation prediction value indicating the (k+1) th time of the monitoring object, m indicating the number of the fuzzy subsets,weights representing the j-th said fuzzy subset, j e [1, m]The k time is the current time, n represents n continuous times before the current time; Represents the jth placeA state prediction result, θ, of the fuzzy subset for the (k+1) th time of the monitored object j 、p ji For optimizing parameters of independent prediction model, x' i Representing the ith said sequential array value, y' i And the state evaluation value corresponding to the ith time sequence number series value is represented.
Preferably, the membership function model is:
wherein,respectively representing the kth moment and the (k-1) moment, wherein the ith time sequence data in the input data set and the corresponding jth membership function value of the fuzzy subset; x represents a time sequence number series value and a state evaluation value at a current time and a plurality of previous times; /> Are all optimization parameters of the membership function model.
Preferably, in step S121, the formula is as follows:
calculating the adaptability of each fuzzy subset and the corresponding fuzzy rule; wherein w is j Representing the adaptability of the j-th fuzzy subset to which the time sequence data belongs and the corresponding fuzzy rule, (2n+2) representing (n+1) time sequence values and corresponding (n+1) state evaluation values of the input data set corresponding to the fuzzy subset,and (3) representing membership function values of the ith time sequence data and the jth fuzzy subset corresponding to the ith time sequence data, wherein the values of i, j and n are natural numbers.
Preferably, in step S131, the formula is as follows:
and carrying out normalization processing on the fitness of each fuzzy subset to obtain the weight of each fuzzy subset.
Preferably, the early warning method of the intelligent gateway further comprises the following steps: training an intelligent early warning model based on a fuzzy neural network algorithm based on an adam optimization algorithm according to a pre-acquired training set, the fuzzy subset and the fuzzy rule to obtain the intelligent early warning model after training, wherein the intelligent early warning model specifically comprises the following components: based on a K-means method, obtaining a clustering center of the training set to divide the training set into a plurality of fuzzy subsets; each fuzzy subset corresponds to a preset fuzzy rule; and optimizing parameters of the intelligent early warning model according to the clustering center of the training set, the fuzzy subset and the fuzzy rule based on an adam optimization algorithm to obtain the intelligent early warning model after training is completed.
Preferably, the K-means based method acquires a cluster center of the training set to divide the training set into a plurality of fuzzy subsets, including: randomly selecting m time sequence data from the training set as an initial clustering center of the training set; wherein m is a natural number; the training set comprises a plurality of sample time sequence data, each sample time sequence data comprises a sample time sequence number sequence value of the monitoring object and a sample state evaluation value corresponding to the sample time sequence number sequence value, and the sample state evaluation value is obtained by manual labeling; dividing the training set into m classes according to the distances from each sample time sequence data in the training set to m initial clustering centers respectively, and according to a formula:
Calculating a cluster center of each class; wherein, c t Representing the cluster center of the t-th class, t epsilon (0, m), and x' represents the sample time sequence data; according to each sample time sequence data in the training set, the training set sends the training set to the clustering center c t Re-dividing the training set into m classes; performing loop iteration on the division of the training set until the clustering center c t No longer varying, resulting in m of said fuzzy subsets.
Preferably, the optimization algorithm based on adam optimizes parameters of the intelligent early-warning model according to the clustering center of the training set, the fuzzy subset and the fuzzy rule to obtain the trained intelligent early-warning model, which specifically comprises: an AdamOptimezer optimizer based on a TensorFlow computing framework carries out iterative optimization on parameters of the intelligent early warning model according to a preset objective function according to the training set, the fuzzy subset and the fuzzy rule to obtain the intelligent early warning model after training is completed; wherein the objective function is:
wherein y' k+1 A sample state evaluation prediction value indicating the (k+1) th time of the monitoring object,a sample expected predicted value indicating the (k+1) th time of the monitoring object.
The embodiment of the application also provides an early warning system of an intelligent gateway, wherein an intelligent early warning model based on a fuzzy neural network algorithm is deployed in the intelligent gateway, and the early warning system comprises: the evaluation unit is configured to obtain a state evaluation predicted value of the monitoring object at the next moment according to a fuzzy subset, a preset fuzzy rule corresponding to the fuzzy subset and the acquired input data set of the monitoring object based on the intelligent early warning model after training; the state evaluation predicted value is used for representing the prediction of the running state of the monitoring object at the next moment; the fuzzy subset is obtained by dividing a training set obtained in advance through a clustering method; the training set is used for training the intelligent early warning model, the input data set comprises time sequence data of the current moment and a plurality of previous moments, and each time sequence data comprises a time sequence value of the monitoring object and a state evaluation value corresponding to the time sequence value; and the early warning unit is configured to compare the state evaluation predicted value with a preset state threshold value, and early warn the running state of the monitored object at the next moment in response to the state evaluation predicted value being larger than the preset threshold value.
Preferably, the evaluation unit includes: a membership subunit configured to calculate a membership function value of each time series data and the fuzzy subset corresponding to the time series data according to a membership function of the fuzzy subset corresponding to the acquired input data set of the monitoring object; an fitness subunit configured to calculate, according to membership function values of each of the time series data and the fuzzy subsets corresponding to the time series data, fitness of each of the fuzzy subsets and the fuzzy rules corresponding to the fuzzy subsets; the normalization subunit is configured to normalize the fitness of each fuzzy subset to obtain the weight of each fuzzy subset; a prediction subunit configured to predict a result of the state prediction of each fuzzy subset and a weight of each fuzzy subset according to a formula:
obtaining a state evaluation predicted value of the monitoring object at the next moment; wherein, according to the independent prediction model:
obtaining each of the resultsThe fuzzy subset predicts the state of the monitored object at the next moment; wherein y is k+1 A state evaluation prediction value indicating the (k+1) th time of the monitoring object, m indicating the number of the fuzzy subsets,weights representing the j-th said fuzzy subset, j e [1, m ]The k time is the current time, n represents n continuous times before the current time;representing the state prediction result of the jth fuzzy subset on the (k+1) th moment of the monitored object, theta j 、p ji For optimizing parameters of independent prediction model, x' i Representing the ith said sequential array value, y' i And the state evaluation value corresponding to the ith time sequence number series value is represented.
Compared with the closest prior art, the technical scheme of the embodiment of the application has the following beneficial effects:
according to the technical scheme, based on an intelligent early warning model deployed in an intelligent gateway, according to time sequence data of a plurality of monitoring objects obtained by data acquisition of a plurality of monitoring objects by a plurality of data acquisition modules distributed in the intelligent gateway, the running state of the monitoring objects at the next moment is predicted, and a state evaluation predicted value of the monitoring objects at the next moment is obtained; then, the state evaluation predicted value of the next moment of the monitored object is compared with a preset state threshold value, so that the running state of the next moment of the monitored object is early-warned; and when the state evaluation predicted value is larger than a preset state threshold value, early warning is carried out on the running state of the monitored object at the next moment. Therefore, the method and the device realize the active prediction of the fault condition of the monitored object and avoid the fault in time.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute an undue limitation to the application. Wherein:
fig. 1 is a schematic structural diagram of a distributed industrial data acquisition early warning intelligent gateway according to some embodiments of the present application;
FIG. 2 is a schematic diagram of a core data processing unit according to some embodiments of the disclosure;
FIG. 3 is a schematic diagram of a data acquisition unit according to some embodiments of the disclosure;
fig. 4 is a flow chart of an early warning method of an intelligent gateway according to some embodiments of the present application;
fig. 5 is a flowchart of step S101 in an early warning method of an intelligent gateway according to some embodiments of the present application;
fig. 6 is a schematic structural diagram of an early warning system of an intelligent gateway according to some embodiments of the present application;
fig. 7 is a schematic structural diagram of an evaluation unit according to some embodiments of the present application.
Reference numerals illustrate:
100-a core data processing module; 200-a data acquisition module; 300-monitoring a subject; 400-target terminal;
101-a first microcontroller; 102-a first LoRa communication sub-module; 103-a telecommunications sub-module; 104-a memory module; 105-a first power module;
201-a second microcontroller; 202-a second LoRa communication sub-module; 203-a standard interface module; 204-a second power module;
301-an evaluation unit; 302-an early warning unit; 311-membership subunit; 321-fitness subunit; 331-normalization subunit; 341-predictor unit.
Detailed Description
The present application will be described in detail below with reference to the accompanying drawings in conjunction with embodiments. Various examples are provided by way of explanation of the present application and not limitation of the present application. Indeed, it will be apparent to those skilled in the art that modifications and variations can be made in the present application without departing from the scope or spirit of the application. For example, features illustrated or described as part of one embodiment can be used on another embodiment to yield still a further embodiment. Accordingly, it is intended that the present application include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Fig. 1 is a schematic structural diagram of a distributed industrial data acquisition early warning intelligent gateway according to some embodiments of the present application; FIG. 2 is a schematic diagram of a core data processing unit according to some embodiments of the disclosure; FIG. 3 is a schematic diagram of a data acquisition unit according to some embodiments of the disclosure; as shown in fig. 1-3, the distributed industrial data acquisition early warning intelligent gateway includes: the system comprises a plurality of data acquisition modules 200 and a core data processing module 100, wherein the plurality of data acquisition modules 200 are distributed, each data acquisition module 200 can acquire time sequence data of a monitoring object 300, the plurality of monitoring objects 300 are arranged, the plurality of monitoring objects 300 correspond to the plurality of data acquisition modules 200, and the plurality of monitoring objects 300 have different data transmission protocols; the data acquisition module 200 is internally provided with a multi-protocol analysis sub-module, and the multi-protocol analysis sub-module can analyze the industrial data transmission protocol of the monitored object 300 so as to analyze the acquired time sequence data into a format of the transmission marking data to be tested of the message queue; the core data processing module 100 is in communication connection with each data acquisition module 200 through a radio technology, and analyzes time sequence data in a telemetry transmission standard data format in a message queue analyzed by the received data acquisition modules 200 based on a preset intelligent early warning module to determine whether early warning information is generated.
In the embodiment of the present application, the data acquisition module 200 can acquire time series data of the monitoring objects 300, and a plurality of monitoring objects 300 correspond to a plurality of data acquisition modules 200. The plurality of data acquisition modules 200 and the plurality of monitoring objects 300 may have a one-to-one or one-to-many relationship, that is, one data acquisition module 200 may monitor one monitoring object 300, or may monitor a plurality of monitoring objects 300 at the same time.
In the embodiment of the present application, a distributed deployment mode of a master multi-slave is adopted between the plurality of data acquisition modules 200 and the core data processing module 100, the multi-protocol analysis sub-module built in the data acquisition module 200 analyzes the industrial data transmission protocol of the monitoring object 300, analyzes the acquired time sequence data of the monitoring object 300 into a message queue telemetry transmission (Message Queuing Telemetry Transport, abbreviated as MQTT) standard data format, and sends the message queue telemetry transmission (Message Queuing Telemetry Transport, abbreviated as MQTT) standard data format to the core data transmission module, thereby realizing unified access of multiple different monitoring objects 300 and eliminating data islands among different monitoring objects 300.
In this embodiment of the present application, the multi-protocol parsing sub-module may be configured to parse common industrial data transmission protocols such as modbus, opcua, profinet, etc., and the data acquisition module 200 performs uplink and downlink communications with the core data processing module 100 according to the standard data format of MQTT. Specifically, the data acquisition module 200 and the core data processing module 100 are both provided with a LoRa communication sub-module adopting a Radio technology (LoRa for short), so that uplink and downlink communication between the data acquisition module 200 and the core data processing module 100 is realized, and ultra-Long distance and high sensitivity communication between the data acquisition module 200 and the core data processing module 100 is ensured. Further, the distance between the core data processing module 100 and the data acquisition module 200 through the radio technology is greater than 11.5 km; the communication sensitivity between the core data processing module 100 and the data acquisition module 200 is not lower than-140 decibel milliwatt (dbm).
In this embodiment, the data acquisition module 200 is provided with a plurality of communication standard interfaces, and the communication standard interfaces can be connected with a sensor, where the sensor is used for acquiring time sequence data of the monitored object 300. Further, the data acquisition module 200 communicates with the core data processing module 100 through the LoRa network, and the data acquisition module 200 is provided with a plurality of ethernet interfaces; specifically, the data acquisition module 200 includes 2 ethernet interfaces and 2 RS485/RS232 communication standard interfaces, where the data acquisition module 200 can communicate with other programmable logic controllers, sensors, etc. through the communication standard interfaces (ethernet interfaces, RS485/RS 232) to extend the distributed industrial data acquisition early warning intelligent gateway.
In this embodiment of the present application, the core data processing module 100 is internally provided with a remote communication sub-module 103, and the remote communication sub-module 103 can transmit the time sequence data and the early warning information in the message queue telemetry transmission standard data format to the target terminal 400, so that the target terminal 400 can display the time sequence data and the early warning information in the message queue telemetry transmission standard data format in real time. Therefore, the time sequence data acquired by the data acquisition unit and the generated early warning information can be effectively ensured to be timely transmitted to the target terminal 400, and accurate countermeasures can be timely taken by the target terminal 400.
Further, a wireless communication unit is built in the remote communication sub-module 103, and the wireless communication unit can send the time sequence data and the early warning information of the message queue telemetry transmission standard data format of the target terminal 400 to the target user in real time in a short message form so as to be checked by the target user. Therefore, the real-time performance of the time sequence data and the early warning information of different monitoring objects 300 is further improved, so that a target user can timely and accurately master the working state of each monitoring object 300, different adjustment can be carried out on different monitoring objects 300, and the efficient and safe work of each monitoring object 300 is ensured.
In this application implementation, distributed industrial data acquisition early warning intelligent gateway still includes: the power module is electrically connected with the data acquisition module 200 and the core data module respectively. Further, the intelligent gateway for distributed industrial data acquisition and early warning further comprises: the storage module 104 is in communication connection with the core data processing module 100 to store time sequence data and early warning information in a message queue telemetry transmission standard data format; and the memory module 104 is electrically connected to the power module. Therefore, the high-efficiency and safe work of the data acquisition unit, the core data processing unit and the storage unit is effectively ensured, and the reliability of the distributed industrial data acquisition early-warning intelligent gateway is improved.
In the embodiment of the present application, the wireless communication sleep current between the data acquisition module 200 and the core data processing module 100 is less than 1.8 microamps. Thereby, the energy consumption can be effectively reduced.
In the embodiment of the present application, the core data processing unit is provided with a first microcontroller 101, a first LoRa communication sub-module 102, a remote communication sub-module 103, a storage module 104 and a first power module 105; the data acquisition unit is provided with a second microcontroller 201, a second LoRa communication sub-module 202, a standard interface module 203 and a second power module 204. The first power module 105 and the second power module 204 may be the same power module, and provide dc_12v voltage for the core data processing unit and the data acquisition unit.
In this embodiment, the standard interface module 203 includes 2 ethernet interfaces (RJ 45) and 2 RS485/RS232 communication standard interfaces, and is connected to the sensor through the RS485/RS232 communication standard interfaces, and the sensor collects the time sequence data of the monitored object 300.
The first microcontroller 101 is preset with an intelligent early warning model, and analyzes the time sequence data of the message queue telemetering transmission standard data format analyzed by the received data acquisition module 200 to determine whether early warning information is generated; the second microcontroller 201 is preset with a multi-protocol analysis sub-module, which can analyze the industrial data transmission protocol of the monitored object 300 to analyze the acquired time sequence data into a message queue telemetry transmission standard data format.
The first LoRa communication sub-module 102 is in communication connection with the second LoRa communication sub-module 202, so that the core data processing unit plays a role of a LoRa gateway, and receives/transmits data to the data acquisition module 200, thereby ensuring long-distance and high-sensitivity communication between the first microcontroller 101 and the second microcontroller 201.
The remote communication sub-module 103 is connected with the first microcontroller 101, is internally provided with a wireless communication unit, and transmits time sequence data and early warning information in a standard data format of the telemetry transmission of the message queue of the target terminal 400 to the target user in real time in a short message form so as to be checked by the target user; the remote communication sub-module 103 is further provided with an ethernet interface, so that the data transmission and communication capability between the data acquisition module 200 and other devices distributed and deployed such as programmable logic controllers and sensors are enhanced by connecting with the ethernet interface in the data acquisition module 200.
The storage module 104 is connected with the first microcontroller 101 and can store time sequence data and early warning information in a message queue telemetry transmission standard data format.
In the embodiment of the application, the first microcontroller 101 adopts a intel haswell Soc architecture chip and is matched with an ultra-low voltage Core i3-4010U processor, and the processor is provided with a Graphics HD 5000 graphic chip with powerful functions, and the design power consumption is only 15 watts, so that the excellent balance of performance, power consumption and volume is achieved. Wherein, the dimension 120mm x 120mm Nano-ITX of the main board of the Core i3-4010U processor has no passive heat dissipation of a fan, can run at a wide temperature, has a working temperature of-20-70 ℃, has a mechanical vibration applicable frequency of 10-150 Hz and has an acceleration amplitude of 24m/s. The second microcontroller 201 employs a black radiation-resistant electrolytic lead-zinc-plated steel plate (Steel, electrodeposition, cold, common, abbreviated as SECC) including based on -STM 32 microprocessor of M3.
In the embodiment of the application, the first microcontroller 101, the second microcontroller 201 and the LoRa wireless networking all adopt low power consumption, the power consumption of the first microcontroller 101 and the second microcontroller 201 is not more than 15w, the LoRa dormancy current is less than 1.8 μa, and the system energy consumption is optimized to the greatest extent.
In the implementation of the application, the data acquisition unit is used for acquiring data of the sensors and the monitoring objects 300 with various protocols and standardizing the data into the MQTT protocol format, the data are uploaded to the core data processing unit through the LoRa network, the core data processing unit can also transmit the control command to the data acquisition module 200 in the MQTT protocol format through the LoRa network, the data acquisition module 200 is used for carrying out data protocol conversion, and the standard interface module 203 is used for transmitting the control command to the field sensors and the controllers so as to realize the monitoring and control of the monitoring objects 300 or the production process.
The core data processing unit analyzes the data uploaded by the data acquisition unit through the fuzzy neural network anomaly prediction model, judges the anomaly condition and the change trend of the time sequence data, and early warns the necessary anomaly condition in advance, so that the problems of one-sided performance, hysteresis and false alarm caused by a conventional threshold value alarm mode are avoided. Specifically, based on the intelligent early warning model of the fuzzy neural network algorithm, the time sequence data of the message queue telemetering transmission standard data format analyzed by the received data acquisition module 200 is analyzed to determine whether to produce early warning information or not, so that the intelligent early warning of abnormal data is realized. The early warning information flow and the acquisition data flow can be uploaded to the central server platform through a wireless module U8300w arranged in the core data processing unit, and the early warning information can be sent to related personnel through a short message sending function of the wireless module U8300 w. The wireless module U8300w is compatible with 2g,3g,4g networks.
In the embodiment of the present application, an intelligent early warning model and a plurality of data acquisition modules 200 based on a fuzzy neural network algorithm are distributed in an intelligent gateway, and the plurality of data acquisition modules 200 are distributed in the intelligent gateway and are used for acquiring data of a plurality of monitoring objects 300 to obtain a plurality of time sequence data of the plurality of monitoring objects 300, and the intelligent early warning model is used for predicting the running state of the monitoring objects 300 according to the time sequence data.
In the embodiment of the present application, the time sequence data set (i.e. the input data set) of a certain monitored variable of the monitored object 300 acquired by the data acquisition module 200 is D = { (x' 1 ，y′ 1 )，(x′ 2 ，y′ 2 )，(x′ 3 ，y′ 3 )…(x′ n ，y′ n ) (x ', y ') represents time series data of the monitoring object 300 acquired by the data acquisition module 200, x ' represents time series values of the time series data, and y ' represents state evaluation values corresponding to the time series values x '; y' e [0,1]Where 0 indicates that the value is normal, and the operation state of the monitoring object 300 is normal, and the closer the value of y is to 1, the more serious the abnormality degree of the operation state of the monitoring object 300 is.
In the examples of the present application, x= { X 'is defined' 1 ，x′ 2 ，x′ 3 …x′ n A time series value representing the time of this monitored variable 1 to n, defining y= { Y' 1 ，y′ 2 ，y′ 3 …y′ n Time series value x= { X 'representing time series from 1 to n times of the monitored variable' 1 ，x′ 2 ，x′ 3 …x′ n A state evaluation value corresponding to the first state;
as shown in fig. 4, the early warning method of the intelligent gateway includes:
step S101, based on the intelligent early warning model after training, obtaining a state evaluation predicted value of the monitored object 300 at the next moment according to the fuzzy subset, a preset fuzzy rule corresponding to the fuzzy subset and the acquired input data set of the monitored object 300; wherein, the state evaluation predicted value is used for representing the prediction of the running state of the monitored object 300 at the next moment; the fuzzy subset is obtained by dividing a training set obtained in advance through a clustering method; the training set is used for training the intelligent early warning model, the input data set comprises time sequence data of the current moment and a plurality of previous moments, and each time sequence data comprises a time sequence number value of a monitoring object and a state evaluation value corresponding to the time sequence number value;
specifically, a pre-acquired training set is divided based on a K-means method to obtain fuzzy subsets.
In the embodiment of the present application, the data acquisition module 200 can acquire time series data of the monitoring objects 300, and a plurality of monitoring objects 300 correspond to a plurality of data acquisition modules 200. The plurality of data acquisition modules 200 and the plurality of monitoring objects 300 may have a one-to-one or one-to-many relationship, that is, one data acquisition module 200 may monitor one monitoring object 300, or may monitor a plurality of monitoring objects 300 at the same time. It should be noted that, the data acquisition module 200 performs data acquisition on the monitored object 300 to obtain a time sequence number sequence value of the monitored object 300, the state monitoring value corresponding to the time sequence number sequence value at the current time is obtained by predicting the trained intelligent early warning model, and when the state evaluation value at the next time of the monitored object is predicted, the state evaluation predicted value corresponding to the current time is the state evaluation value at the current time.
The data acquisition module 200 is internally provided with a multi-protocol analysis sub-module for analyzing industrial data transmission protocols of a plurality of monitoring objects 300 with different data transmission protocols, analyzing the acquired time sequence data of the monitoring objects 300 into a message queue telemetry transmission (Message Queuing Telemetry Transport, abbreviated as MQTT) standard data format, realizing unified access of a plurality of different monitoring objects 300 and eliminating data islands among the different monitoring objects 300.
In some alternative embodiments, the partitioning of the pre-acquired training set by the K-means method to obtain the fuzzy subset is specifically: based on the K-means method, a clustering center of the training set is obtained to divide the training set into a plurality of fuzzy subsets.
In the embodiment of the present application, first, m (m is a natural number) sample time sequence data are randomly selected from a training set obtained by performing time sequence data acquisition on a monitored object 300 by a data acquisition module 200 as an initial clustering center of the training set; and calculating the distance from each sample time sequence data in the training set to m initial clustering centers, dividing the training set into m classes according to the distance from each sample time sequence data in the training set to m initial clustering centers, and calculating the clustering centers of each class according to a formula (1). The formula (1) is as follows:
Wherein, c t Representing the cluster center of the t-th class, t E (0, m), and x' represents sample time series data. Here, it should be noted that each sample timing data is a vector whose elements are a sample timing sequence value and a sample state evaluation value of the sample timing data, that is, the vector is expressed in terms of (sample timing sequence value, sample state evaluation value).
In this embodiment of the present application, each sample time sequence data includes a sample time sequence number sequence value of the monitoring object and a sample state evaluation value corresponding to the sample time sequence number sequence value, where the sample state evaluation value is obtained by manual labeling. For the training set divided into m classes, recalculating time sequence data of each sample in the training set to a clustering center c t Re-dividing the sample time series data set into m classes; sequentially, carrying out loop iteration on the division of the training set until a clustering center c t No longer varying, resulting in m fuzzy subsets.
Fig. 5 is a flowchart of step S101 in an early warning method of an intelligent gateway according to some embodiments of the present application; as shown in fig. 5, based on the trained intelligent early warning model, according to the fuzzy subset and the preset fuzzy rule corresponding to the fuzzy subset, and the acquired input data set of the monitored object 300, obtaining the state evaluation prediction value of the monitored object 300 at the next moment includes:
Step S111, calculating membership function values of each time sequence data and the corresponding fuzzy subset according to the membership function model of the fuzzy subset corresponding to the acquired input data set of the monitoring object;
in this embodiment of the present application, each fuzzy subset corresponding to a component (a time sequence number or a state evaluation value) of the input time sequence data has a membership function, and according to the corresponding membership function, the membership function value of each fuzzy subset corresponding to the time sequence data is calculated. Here, the membership function value characterizes the degree of similarity of the component (time series value or state evaluation value) of the input time series data to its corresponding fuzzy subset.
Specifically, the membership function model is shown in formula (2):
in the method, in the process of the invention,respectively representing the kth moment and the (k-1) moment, wherein the ith time sequence data and the corresponding jth fuzzy subset membership function value; x represents a time sequence number series value and a state evaluation value at a current time and a plurality of previous times; />Are all of membership degreeOptimization parameters of the function model. Here, a->And the optimization parameters in the membership function are obtained after the intelligent early warning model based on the fuzzy neural network is trained. Wherein (1) >Representation ofFeedback coefficient of>Training->Is 0.
Step S121, calculating the fitness of each fuzzy subset and the corresponding fuzzy rule according to a formula (3) according to the membership function value of each time sequence data and the corresponding fuzzy subset; equation (3) is as follows:
wherein w is j Represents the adaptability of the j-th fuzzy subset to which the time sequence data belongs and the corresponding fuzzy rule, (2n+2) represents the (n+1) time sequence number sequence values of the input data set corresponding to the fuzzy subset and the corresponding (n+1) state evaluation values thereof,and the membership function value of the ith time sequence data and the jth fuzzy subset corresponding to the ith time sequence data is represented, and the values of i, j and n are natural numbers.
In the embodiment of the application, the adaptability of each fuzzy subset and the corresponding fuzzy rule characterizes the similarity degree of the input time sequence data and the corresponding fuzzy rule. The higher the fitness is, the higher the similarity between the inputted time series data and the fuzzy rule is.
Step S131, carrying out normalization processing on the fitness of each fuzzy subset to obtain the weight of each fuzzy subset;
specifically, the fitness of each fuzzy subset is normalized according to the formula (4) to obtain the weight of each fuzzy subset. Equation (4) is as follows:
In the method, in the process of the invention,weights representing the j-th fuzzy subset, j.epsilon.1, m]。
In the embodiment of the application, the weight of each fuzzy subset represents the specific gravity of the fuzzy subset when the running state of the monitored object 300 at the next moment is predicted through the intelligent early warning model.
Step S141, calculating a state evaluation predicted value of the monitored object 300 at the next moment according to a formula (5) according to the state predicted result of each fuzzy subset and the weight of each fuzzy subset; equation (5) is as follows:
wherein, according to the independent prediction model (as shown in formula (6):
and obtaining a state prediction result of each fuzzy subset on the next moment of the monitored object 300. Wherein y is k+1 A state evaluation prediction value representing the (k+1) th time of the monitoring object 300, m represents the number of fuzzy subsets,represents the j thWeights of fuzzy subsets, j.epsilon.1, m]The k time is the current time, n represents n continuous times before the current time; />Representing the state prediction result, θ, of the jth fuzzy subset with respect to the (k+1) th time of the monitored object 300 j 、p ji For optimizing parameters of independent prediction model, x' i Represents the ith time series value, y' i The state evaluation value corresponding to the i-th time series value is represented.
In the embodiment of the application, the intelligent early warning model is composed of a front part and a back part, wherein the front part is composed of an input layer, a language variable layer, an fitness calculation layer and a normalization operation layer, and the back part is composed of m sub-networks (independent prediction models ) And the parallel components. Receiving the collected time sequence data of the monitoring object 300 at an input layer; adopting a recursive network structure at a language variable layer, and calculating membership function values of components of time sequence data input in an input layer and corresponding fuzzy subsets; in the fitness calculation layer, each node represents a fuzzy rule, and the fitness of each rule, namely the fitness of each fuzzy subset and the corresponding fuzzy rule, is calculated through a formula (3); at the normalization operation layer, the weights of each fuzzy subset are calculated. An independent prediction model is deployed in the back part, a state prediction result of each fuzzy subset on the next moment of the monitored object 300 is calculated, and then the state prediction result of each fuzzy subset on the next moment of the monitored object 300 is weighted and summed according to the weight of the corresponding fuzzy subset, so that a state evaluation prediction value of the next moment of the monitored object 300 is obtained.
In the embodiment of the present application, the node in the linguistic variable layer is a calculation unit of the fitness of each fuzzy subset, and the fuzzy rule is expressed as follows:
R j ：if x k-n isand x k-n+1 is/>...，x k is/>
and y k-n isand y k-n+1 is/>...，y k is/>
then
step S102, comparing the state evaluation predicted value with a preset state threshold, and early warning the running state of the monitored object 300 at the next moment in response to the state evaluation predicted value being greater than the preset state threshold.
In the embodiment of the present application, the preset state threshold is 0.5, and when the state evaluation predicted value of the monitored object 300 at the next moment is greater than or equal to 0.5, the running state of the monitored object 300 at the next moment is pre-warned.
In the embodiment of the application, based on an intelligent early warning model deployed in an intelligent gateway, according to time sequence data of a monitoring object 300 obtained by data acquisition of a plurality of monitoring objects 300 by a plurality of data acquisition modules 200 distributed in the intelligent gateway, the running state of the monitoring object 300 at the next moment is predicted, and a state evaluation predicted value of the monitoring object 300 at the next moment is obtained; then, the state evaluation predicted value of the next moment of the monitored object 300 is compared with a preset state threshold value, so that the running state of the next moment of the monitored object 300 is early-warned; and when the state evaluation predicted value is larger than the preset state threshold value, the running state of the monitored object 300 at the next moment is pre-warned. Thus, the active prediction of the fault condition of the monitored object 300 is realized, and the fault is avoided in time.
In some optional embodiments, the early warning method of the intelligent gateway further includes: according to a pre-acquired training set, a fuzzy subset and a fuzzy rule, training an intelligent early warning model based on a fuzzy neural network algorithm based on an adam optimization algorithm to obtain a trained intelligent early warning model, wherein the intelligent early warning model comprises the following specific steps:
Firstly, acquiring a clustering center of a training set based on a K-means method to divide the training set into a plurality of fuzzy subsets; each fuzzy subset corresponds to a preset fuzzy rule;
in the embodiment of the application, randomly selecting m sample time sequence data from a training set as an initial clustering center of the training set; wherein m is a natural number; the training set comprises a plurality of sample time sequence data; dividing the training set into m classes according to the distances from each sample time sequence data in the training set to m initial clustering centers respectively, and calculating the clustering centers of each class according to a formula (1); according to each sample time sequence data in the training set to the clustering center c t Re-dividing the training set into m classes; performing loop iteration on the division of the training set until the clustering center c t No longer varying, resulting in m fuzzy subsets.
Secondly, performing iterative optimization on parameters of the intelligent early warning model according to a training set, a fuzzy subset and a fuzzy rule and a preset objective function to obtain the trained intelligent early warning model based on an AdamOptimezer of a TensorFlow computing framework; wherein, the objective function (as shown in formula (7)) is:
wherein y' k+1 A sample state evaluation prediction value indicating the (k+1) th time of the monitoring object 300, A sample expected predicted value at the (k+1) th time of the monitoring object 300 is represented. Wherein y' k+1 Can be obtained according to the above formula (5).
In the embodiment of the application, performing iterative optimization on parameters of the intelligent early-warning model according to the objective function refers to continuously optimizing parameters v, c and b in the membership function model and parameters theta in the independent prediction model through training of the intelligent early-warning model j 、p ji Until the value of the objective function E (k+1) approaches 0, to obtainθ j 、p ji The training of the intelligent early warning model is completed.
In the embodiment of the application, in the training process of the intelligent early warning model,θ j 、p ji the initial value of (1) is a random value, and an AdamOptimezer optimizer adopting a TensorFlow computing framework is repeatedly called in the iteration process to carry out iteration optimization. The maximum number of iterative optimization is 1000, that is, after the intelligent early warning model is circularly iterated 1000 times, the value of the objective function E (k+1) is still not approaching to 0, and the parameter value of the last iteration in sequence is determined as +.>θ j 、p ji Is used for the optimization of the values of (a).
Fig. 6 is a schematic structural diagram of an early warning system of an intelligent gateway according to some embodiments of the present application; as shown in fig. 6, an intelligent early warning model early warning system based on a fuzzy neural network algorithm is deployed in an intelligent gateway, and the intelligent early warning model early warning system comprises: an evaluation unit 301 and an early warning unit 302. The evaluation unit 301 is configured to obtain a state evaluation predicted value of the monitored object 300 at the next moment according to the fuzzy subset, a preset fuzzy rule corresponding to the fuzzy subset, and the acquired input data set of the monitored object 300 based on the trained intelligent early warning model; wherein, the state evaluation predicted value is used for representing the prediction of the running state of the monitored object 300 at the next moment; the fuzzy subset is obtained by dividing a training set obtained in advance through a clustering method; the training set is used for training the intelligent early warning model, the input data set comprises time sequence data of the current moment and a plurality of previous moments, and each time sequence data comprises a time sequence number sequence value of the monitoring object and the state evaluation value corresponding to the time sequence number sequence value; the early warning unit 302 is configured to compare the state evaluation prediction value with a preset state threshold value, and early warn the operation state of the monitored object 300 at the next moment in response to the state evaluation prediction value being greater than the preset threshold value.
In this embodiment of the present application, a plurality of data acquisition modules 200 are distributed in the intelligent gateway, where the data acquisition modules 200 are configured to perform data acquisition on a plurality of monitored objects 300, so as to obtain a plurality of time sequence data of the plurality of monitored objects 300.
FIG. 7 is a schematic diagram of an evaluation unit provided according to some embodiments of the present application; as shown in fig. 7, the evaluation unit 301 includes: membership subunit 311, fitness subunit 322, normalization subunit 331, and prediction subunit 341. A membership subunit 311 configured to calculate a membership function value of each time sequence data and the fuzzy subset corresponding to the time sequence data according to the membership function of the fuzzy subset corresponding to the acquired input data set of the monitoring object; an fitness subunit 322 configured to calculate, according to the membership function value of each time-series data and its corresponding fuzzy subset, the fitness of each fuzzy subset and its corresponding fuzzy rule; the normalization subunit 331 is configured to normalize the fitness of each fuzzy subset to obtain the weight of each fuzzy subset; the prediction subunit 341 is configured to, according to the state prediction result of each fuzzy subset and the weight of each fuzzy subset, and according to the formula:
Obtaining a state evaluation predicted value of the monitored object 300 at the next moment; wherein, according to the independent prediction model:
obtaining a state prediction result of each fuzzy subset on the next moment of the monitored object 300; wherein y is k+1 A state evaluation prediction value representing the (k+1) th time of the monitoring object 300, m represents the number of fuzzy subsets,weights representing the j-th fuzzy subset, j.epsilon.1, m]The k time is the current time, n represents n continuous times before the current time; />Representing the state prediction result, θ, of the jth fuzzy subset with respect to the (k+1) th time of the monitored object 300 j 、p ji For optimizing parameters of independent prediction model, x' i Represents the ith time series value, y' i The state evaluation value corresponding to the i-th time series value is represented.
The early warning system of the intelligent gateway provided by the embodiment of the application can realize the flow and the steps of the early warning method of the intelligent gateway of any embodiment, and achieve the same beneficial effects, and are not described in detail herein.
The foregoing description is only of the preferred embodiments of the present application and is not intended to limit the same, but rather, various modifications and variations may be made by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.
Claims (7)
1. An intelligent gateway early warning method is characterized in that an intelligent early warning model based on a fuzzy neural network algorithm is deployed in the intelligent gateway, and the early warning method comprises the following steps:
step S101, based on the intelligent early warning model after training, obtaining a state evaluation predicted value of the monitoring object at the next moment according to a fuzzy subset, a preset fuzzy rule corresponding to the fuzzy subset and an acquired input data set of the monitoring object;
the state evaluation predicted value is used for representing the running state of the monitoring object at the next moment; the fuzzy subset is obtained by dividing a training set obtained in advance through a clustering method; the training set is used for training the intelligent early warning model, the input data set comprises time sequence data of the current moment and a plurality of previous moments, and each time sequence data comprises a time sequence value of the monitoring object and a state evaluation value corresponding to the time sequence value;
step S102, comparing the state evaluation predicted value with a preset state threshold value, and early warning the running state of the monitored object at the next moment in response to the state evaluation predicted value being larger than the preset state threshold value;
The early warning method of the intelligent gateway further comprises the following steps: training an intelligent early warning model based on a fuzzy neural network algorithm based on an adam optimization algorithm according to a pre-acquired training set, the fuzzy subset and the fuzzy rule to obtain the intelligent early warning model after training, wherein the intelligent early warning model specifically comprises the following components:
based on a K-means method, obtaining a clustering center of the training set to divide the training set into a plurality of fuzzy subsets; each fuzzy subset corresponds to a preset fuzzy rule;
optimizing parameters of the intelligent early warning model based on an adam optimization algorithm according to a clustering center of the training set, the fuzzy subset and the fuzzy rule to obtain the intelligent early warning model after training is completed;
the step S101 includes:
step S111, calculating a membership function value of each time sequence data and the corresponding fuzzy subset according to the membership function model of the fuzzy subset corresponding to the acquired input data set of the monitoring object;
step S121, calculating the fitness of each fuzzy subset and the corresponding fuzzy rule according to the membership function value of each time sequence data and the corresponding fuzzy subset;
Step S131, carrying out normalization processing on the fitness of each fuzzy subset to obtain the weight of each fuzzy subset;
step S141, according to the state prediction result of each fuzzy subset and the weight of each fuzzy subset, according to the formula:
obtaining a state evaluation predicted value of the monitoring object at the next moment;
wherein, according to the independent prediction model:
obtaining a state prediction result of each fuzzy subset on the next moment of the monitoring object;
wherein y is k+1 A state evaluation prediction value indicating the (k+1) th time of the monitoring object, m indicating the number of the fuzzy subsets,weights representing the j-th said fuzzy subset, j e [1, m]The k time is the current time, n represents n continuous times before the current time;
representing the state prediction result of the jth fuzzy subset on the (k+1) th moment of the monitored object, theta j 、p ji For optimizing parameters of independent prediction model, x' i Representing the ith said sequential array value, y' i And the state evaluation value corresponding to the ith time sequence number series value is represented.
2. The method for early warning of an intelligent gateway according to claim 1, wherein in step S111, the membership function model is:
Wherein,respectively representing the kth moment and the (k-1) moment, wherein the ith time sequence data in the input data set and the corresponding jth membership function value of the fuzzy subset; x represents a time sequence number series value and a state evaluation value at a current time and a plurality of previous times; /> Are all optimization parameters of the membership function model.
3. The method for early warning of an intelligent gateway according to claim 1, characterized in that in step S121, the following formula is adopted:
calculating the adaptability of each fuzzy subset and the corresponding fuzzy rule; wherein w is j Representing the adaptability of the j-th fuzzy subset to which the time sequence data belongs and the corresponding fuzzy rule, (2n+2) representing (n+1) time sequence values and corresponding (n+1) state evaluation values of the input data set corresponding to the fuzzy subset,membership function representing the ith said time series data and its corresponding jth said fuzzy subsetThe values of i, j and n are natural numbers.
4. The method for early warning of intelligent gateway according to claim 3, characterized in that in step S131, according to the formula:
and carrying out normalization processing on the fitness of each fuzzy subset to obtain the weight of each fuzzy subset.
5. The method for early warning of an intelligent gateway according to any one of claims 1 to 4, characterized in that,
the K-means based method acquires a clustering center of the training set to divide the training set into a plurality of fuzzy subsets, including:
randomly selecting m time sequence data from the training set as an initial clustering center of the training set; wherein m is a natural number; the training set comprises a plurality of sample time sequence data, each sample time sequence data comprises a sample time sequence number sequence value of the monitoring object and a sample state evaluation value corresponding to the sample time sequence number sequence value, and the sample state evaluation value is obtained by manual labeling;
dividing the training set into m classes according to the distances from each sample time sequence data in the training set to m initial clustering centers respectively, and according to a formula:
calculating a cluster center of each class; wherein, c t Representing the cluster center of the t-th class, t epsilon (0, m), and x' represents the sample time sequence data;
according to each sample time sequence data in the training set, the training set sends the training set to the clustering center c t Re-dividing the training set into m classes;
performing loop iteration on the division of the training set until the clustering center c t No longer varying, resulting in m of said fuzzy subsets.
6. The early warning method of any one of claims 1 to 4, wherein the optimization algorithm based on adam optimizes parameters of the intelligent early warning model according to a clustering center of the training set, the fuzzy subset and the fuzzy rule to obtain the trained intelligent early warning model, specifically:
an AdamOptimezer optimizer based on a TensorFlow computing framework carries out iterative optimization on parameters of the intelligent early warning model according to a preset objective function according to the training set, the fuzzy subset and the fuzzy rule to obtain the intelligent early warning model after training is completed;
wherein the objective function is:
wherein y' k+1 A sample state evaluation prediction value indicating the (k+1) th time of the monitoring object,a sample expected predicted value indicating the (k+1) th time of the monitoring object.
7. An intelligent gateway early warning system is characterized in that an intelligent early warning model based on a fuzzy neural network algorithm is deployed in the intelligent gateway, and the early warning system comprises:
the evaluation unit is configured to obtain a state evaluation predicted value of the monitoring object at the next moment according to a fuzzy subset, a preset fuzzy rule corresponding to the fuzzy subset and the acquired input data set of the monitoring object based on the intelligent early warning model after training;
The state evaluation predicted value is used for representing the prediction of the running state of the monitoring object at the next moment; the fuzzy subset is obtained by dividing a training set obtained in advance through a clustering method; the training set is used for training the intelligent early warning model, the input data set comprises time sequence data of the current moment and a plurality of previous moments, and each time sequence data comprises a time sequence value of the monitoring object and a state evaluation value corresponding to the time sequence value;
the early warning unit is configured to compare the state evaluation predicted value with a preset state threshold value, and early warn the running state of the monitored object at the next moment in response to the state evaluation predicted value being larger than the preset state threshold value;
the intelligent early warning model after training is obtained by training the intelligent early warning model based on a fuzzy neural network algorithm based on an adam optimization algorithm according to a pre-acquired training set, the fuzzy subset and the fuzzy rule, and specifically comprises the following steps:
based on a K-means method, obtaining a clustering center of the training set to divide the training set into a plurality of fuzzy subsets; each fuzzy subset corresponds to a preset fuzzy rule;
Optimizing parameters of the intelligent early warning model based on an adam optimization algorithm according to a clustering center of the training set, the fuzzy subset and the fuzzy rule to obtain the intelligent early warning model after training is completed;
the evaluation unit includes:
a membership subunit configured to calculate a membership function value of each time series data and the fuzzy subset corresponding to the time series data according to a membership function of the fuzzy subset corresponding to the acquired input data set of the monitoring object;
an fitness subunit configured to calculate, according to membership function values of each of the time series data and the fuzzy subsets corresponding to the time series data, fitness of each of the fuzzy subsets and the fuzzy rules corresponding to the fuzzy subsets;
the normalization subunit is configured to normalize the fitness of each fuzzy subset to obtain the weight of each fuzzy subset;
a prediction subunit configured to predict a result of the state prediction of each fuzzy subset and a weight of each fuzzy subset according to a formula:
obtaining a state evaluation predicted value of the monitoring object at the next moment;
wherein, according to the independent prediction model:
obtaining a state prediction result of each fuzzy subset on the next moment of the monitoring object;
Wherein y is k+1 A state evaluation prediction value indicating the (k+1) th time of the monitoring object, m indicating the number of the fuzzy subsets,weights representing the j-th said fuzzy subset, j e [1, m]The k time is the current time, n represents n continuous times before the current time;
representing the state prediction result of the jth fuzzy subset on the (k+1) th moment of the monitored object, theta j 、p ji For optimizing parameters of independent prediction model, x' j Representing the ith said sequential array value, y' i And the state evaluation value corresponding to the ith time sequence number series value is represented.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110875021.8A CN113658415B (en) | 2021-07-30 | 2021-07-30 | Early warning method and system of intelligent gateway |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110875021.8A CN113658415B (en) | 2021-07-30 | 2021-07-30 | Early warning method and system of intelligent gateway |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113658415A CN113658415A (en) | 2021-11-16 |
| CN113658415B true CN113658415B (en) | 2024-03-26 |
Family
ID=78478204
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110875021.8A Active CN113658415B (en) | 2021-07-30 | 2021-07-30 | Early warning method and system of intelligent gateway |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113658415B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117081965B (en) * | 2023-10-19 | 2024-01-16 | 山东五棵松电气科技有限公司 | Intranet application load on-line monitoring system |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20080084132A (en) * | 2007-03-15 | 2008-09-19 | 임준식 | Method for extracting nonlinear time series prediction model using neural network with weighted fuzzy membership functions |
| CN103716180A (en) * | 2013-12-04 | 2014-04-09 | 国网上海市电力公司 | Network flow actual forecasting-based network abnormality pre-warning method |
| CN104598984A (en) * | 2014-12-08 | 2015-05-06 | 北京邮电大学 | Fuzzy neural network based fault prediction method |
| CN108645615A (en) * | 2018-04-08 | 2018-10-12 | 太原科技大学 | A kind of Adaptive Fuzzy Neural-network gear method for predicting residual useful life |
| CN109272037A (en) * | 2018-09-17 | 2019-01-25 | 江南大学 | A kind of self-organizing TS pattern paste network modeling method applied to infra red flame identification |
| CN109495296A (en) * | 2018-11-02 | 2019-03-19 | 国网四川省电力公司电力科学研究院 | Intelligent substation communication network state evaluation method based on clustering and neural network |
| CN109977621A (en) * | 2019-04-30 | 2019-07-05 | 西南石油大学 | A kind of air-conditioning failure prediction method based on deep learning |
| CN111142027A (en) * | 2019-12-31 | 2020-05-12 | 国电南瑞南京控制系统有限公司 | Lithium iron phosphate battery state-of-charge monitoring and early warning method based on neural network |
| CN111813084A (en) * | 2020-07-10 | 2020-10-23 | 重庆大学 | Mechanical equipment fault diagnosis method based on deep learning |
| CN111952962A (en) * | 2020-07-30 | 2020-11-17 | 国网江苏省电力有限公司南京供电分公司 | Power distribution network low voltage prediction method based on T-S fuzzy neural network |
| CN112487910A (en) * | 2020-11-24 | 2021-03-12 | 中广核工程有限公司 | Fault early warning method and system for nuclear turbine system |
-
2021
- 2021-07-30 CN CN202110875021.8A patent/CN113658415B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20080084132A (en) * | 2007-03-15 | 2008-09-19 | 임준식 | Method for extracting nonlinear time series prediction model using neural network with weighted fuzzy membership functions |
| CN103716180A (en) * | 2013-12-04 | 2014-04-09 | 国网上海市电力公司 | Network flow actual forecasting-based network abnormality pre-warning method |
| CN104598984A (en) * | 2014-12-08 | 2015-05-06 | 北京邮电大学 | Fuzzy neural network based fault prediction method |
| CN108645615A (en) * | 2018-04-08 | 2018-10-12 | 太原科技大学 | A kind of Adaptive Fuzzy Neural-network gear method for predicting residual useful life |
| CN109272037A (en) * | 2018-09-17 | 2019-01-25 | 江南大学 | A kind of self-organizing TS pattern paste network modeling method applied to infra red flame identification |
| CN109495296A (en) * | 2018-11-02 | 2019-03-19 | 国网四川省电力公司电力科学研究院 | Intelligent substation communication network state evaluation method based on clustering and neural network |
| CN109977621A (en) * | 2019-04-30 | 2019-07-05 | 西南石油大学 | A kind of air-conditioning failure prediction method based on deep learning |
| CN111142027A (en) * | 2019-12-31 | 2020-05-12 | 国电南瑞南京控制系统有限公司 | Lithium iron phosphate battery state-of-charge monitoring and early warning method based on neural network |
| CN111813084A (en) * | 2020-07-10 | 2020-10-23 | 重庆大学 | Mechanical equipment fault diagnosis method based on deep learning |
| CN111952962A (en) * | 2020-07-30 | 2020-11-17 | 国网江苏省电力有限公司南京供电分公司 | Power distribution network low voltage prediction method based on T-S fuzzy neural network |
| CN112487910A (en) * | 2020-11-24 | 2021-03-12 | 中广核工程有限公司 | Fault early warning method and system for nuclear turbine system |
Non-Patent Citations (2)
| Title |
|---|
| 基于模糊神经网络的电能表误差超差风险预测模型;金阳忻;电网技术;全文 * |
| 基于模糊神经网络的网络运行态势感知;付钟杨;中国优秀硕士学位论文全文数据库 信息科技辑;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113658415A (en) | 2021-11-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2020077682A1 (en) | Service quality evaluation model training method and device | |
| CN109872003B (en) | Object state prediction method, object state prediction system, computer device, and storage medium | |
| Kumari et al. | An energy efficient smart metering system using edge computing in LoRa network | |
| CN112684301B (en) | Method and device for detecting power grid faults | |
| CN113658415B (en) | Early warning method and system of intelligent gateway | |
| Zhang et al. | Multi-parameter online measurement IoT system based on BP neural network algorithm | |
| CN116308304A (en) | New energy intelligent operation and maintenance method and system based on meta learning concept drift detection | |
| CN117354171B (en) | Platform health condition early warning method and system based on Internet of things platform | |
| Tang et al. | Energy-efficient sensory data collection based on spatiotemporal correlation in IoT networks | |
| CN220798459U (en) | Distributed industrial data acquisition early warning intelligent gateway | |
| Tang et al. | Mathematical modeling of resource allocation for cognitive radio sensor health monitoring system using coevolutionary quantum-behaved particle swarm optimization | |
| CN112801815B (en) | Power communication network fault early warning method based on federal learning | |
| CN113316085B (en) | Method, device and system for giving alarm for detention in closed space | |
| CN113204915A (en) | PHM design method based on CPS | |
| Zhou et al. | A Novel Wireless Sensor Network Data Aggregation Algorithm Based on Self-organizing Feature Mapping Neutral Network. | |
| CN116682250B (en) | Robot wireless remote control device | |
| Kumar et al. | Efficient anomaly detection methodology for power saving in massive iot architecture | |
| Yadav | Machine Learning in Wireless Sensor Networks: A Survey | |
| CN117692026B (en) | Link sensing method and device for power line communication | |
| CN113885596B (en) | Intelligent monitoring system for sewage treatment | |
| Xu | A New Method for Reliability Evaluation of Wireless Sensor Networks Based on Fuzzy Neural Network. | |
| Almazaideh et al. | A predictive maintenance system for wireless sensor networks: a machine learning approach | |
| Qin et al. | Embedded weather forecast system based on multisource information fusion and perception | |
| CN117240886A (en) | Internet of things equipment management system | |
| Rui | Computer Artificial Intelligence IT Technology Data Transmission Center Construction System Research |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20240221 Address after: 211100 1st floor, building H1, No. 719, Sheng'an Avenue, Jiangning Binjiang Development Zone, Nanjing City, Jiangsu Province Applicant after: JIANGSU ZHANDE MEDICAL ARTICLE CO.,LTD. Country or region after: China Address before: 210012 1st floor, No.2 Yuhua Avenue, Yuhuatai District, Nanjing City, Jiangsu Province Applicant before: Nanjing Vanke Information Technology Co.,Ltd. Country or region before: China |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |