CN116702038A - Cloud edge cooperation-based energy station high-frequency data acquisition and abnormal state identification method - Google Patents

Cloud edge cooperation-based energy station high-frequency data acquisition and abnormal state identification method Download PDF

Info

Publication number
CN116702038A
CN116702038A CN202310687531.1A CN202310687531A CN116702038A CN 116702038 A CN116702038 A CN 116702038A CN 202310687531 A CN202310687531 A CN 202310687531A CN 116702038 A CN116702038 A CN 116702038A
Authority
CN
China
Prior art keywords
frequency data
energy station
data
station equipment
abnormal
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
Application number
CN202310687531.1A
Other languages
Chinese (zh)
Inventor
赵琼
金鹤峰
穆佩红
王炳
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang Yingji Power Technology Co ltd
Original Assignee
Zhejiang Yingji Power Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Zhejiang Yingji Power Technology Co ltd filed Critical Zhejiang Yingji Power Technology Co ltd
Priority to CN202310687531.1A priority Critical patent/CN116702038A/en
Publication of CN116702038A publication Critical patent/CN116702038A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • 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
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • 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
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Probability & Statistics with Applications (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

The invention discloses a cloud-edge cooperation-based energy station high-frequency data acquisition and abnormal state identification method, which comprises the following steps: acquiring vibration, sound and pressure wave data of the energy station equipment by installing a vibration sensor, a sound sensor and a pressure wave sensor at the energy station equipment; the data acquisition module in the edge computing platform acquires the data information of the related energy station equipment to obtain high-frequency data of the energy station equipment; after cluster learning is carried out on the historical high-frequency data of the energy station equipment through the cloud platform, marking the high-frequency data with normal and abnormal data labels, establishing a high-frequency data abnormal detection model, transmitting the high-frequency data to the edge computing platform to detect the acquired real-time high-frequency data of the energy station equipment, and uploading the high-frequency data to the cloud platform if the high-frequency data is detected to be the abnormal high-frequency data; after the cloud platform receives the abnormal high-frequency data of the energy station equipment, the energy station equipment abnormality identification is carried out through a pre-established energy station equipment abnormality identification model, and the abnormal type of the energy station equipment is obtained.

Description

Cloud edge cooperation-based energy station high-frequency data acquisition and abnormal state identification method
Technical Field
The invention belongs to the technical field of comprehensive energy stations, and particularly relates to an energy station high-frequency data acquisition and abnormal state identification method based on cloud edge cooperation.
Background
The comprehensive energy station is built in a comprehensive park or development area, various energy sources (gas, water, steam, heat energy and electric energy) and various energy supply and energy saving equipment (a generator set, a waste heat boiler, a refrigerating unit, a solar panel and the like) are combined, production and energy transportation are controlled and controlled through internet centralized management, an independent energy island is built, and resource integration and comprehensive utilization of regional energy sources are realized.
The internet of things is a fusion application of intelligent perception and recognition technology and pervasive computing and ubiquitous networks, and in the 5G universal internet of things era, the internet of things needs to perceive surrounding objects and physical environments through different types of sensors, so that a basis is provided for data analysis of an application layer of the internet of things. The sensor acquisition system monitors, perceives and acquires information of various environments or monitoring objects in real time, achieves communication among various elements, information spaces and human society in the physical world, has different scales, different acquisition frequencies and different requirements on the sensor acquisition system, and performs high-frequency acquisition on different sensors, so that the sensor acquisition system is required to have large-scale and high-bandwidth real-time transmission, calculation, storage and processing capabilities.
At present, the types of devices involved in the comprehensive energy station are complex, the coupling performance is strong, the normal operation of each device of the energy station is ensured, most of the current operation monitoring of the devices of the energy station usually adopts the technology of the Internet of things to collect and identify the temperature and the flow, few researches are conducted on high-frequency data such as vibration data, sound data and pressure wave data, in addition, the service life of each device of the energy station is long, the time between two faults is long, the requirements of screening-free collection, transmission and storage of the high-frequency data on hardware devices are too high, but the high-frequency data can identify the devices more timely and sharply, the operation analysis and maintenance of the devices of the energy station are helped by operators in advance, and the method is an important guarantee in the unattended scene of the energy station. Therefore, how to optimize the collection mode of the high-frequency data according to the characteristics of the energy station equipment, effectively collect and detect and identify the data, and ensuring the normal and safe operation of the energy station equipment is a problem which needs to be solved at present.
Based on the technical problems, a new cloud-edge-collaboration-based energy station high-frequency data acquisition and abnormal state identification method is needed to be designed.
Disclosure of Invention
The invention aims to solve the technical problem of overcoming the defects of the prior art and providing a cloud-edge-collaboration-based energy station high-frequency data acquisition and abnormal state identification method, wherein an edge computing platform and a cloud platform cooperatively operate, so that the data acquisition and local data detection processing capacity of the edge computing platform can be fully utilized, and the big data processing, model training and data sharing capacity of the cloud platform can be utilized to finally realize the acquisition, abnormal detection and abnormal state identification of high-frequency data.
In order to solve the technical problems, the technical scheme of the invention is as follows:
the invention provides a cloud-edge-collaboration-based energy station high-frequency data acquisition and abnormal state identification method, which comprises the following steps: a high frequency data acquisition stage and an abnormal state identification stage;
the high frequency data acquisition phase comprises the following steps:
acquiring vibration data, sound data and pressure wave data information of the energy station equipment by installing a vibration sensor, a sound sensor and a pressure wave sensor at the energy station equipment;
acquiring related energy station equipment data information according to a predefined energy station equipment acquisition object, high-frequency data acquisition content, high-frequency data acquisition frequency, high-frequency data acquisition start-stop conditions and high-frequency data acquisition priority by a data acquisition module in an edge computing platform to obtain high-frequency data of the energy station equipment; the high-frequency data of the energy station equipment at least comprises sound data, vibration data and pressure wave data of a water pump, a fan, a pipeline and a compressor in the energy station;
The abnormal state identification phase comprises the following steps:
after cluster learning is carried out on the historical high-frequency data of the energy station equipment through the cloud platform, marking normal and abnormal data labels on the high-frequency data, training the historical high-frequency data with the labels, establishing a high-frequency data abnormal detection model, and transmitting the high-frequency data abnormal detection model to the edge computing platform;
detecting the acquired real-time high-frequency data of the energy station equipment through a high-frequency data anomaly detection model in the edge computing platform, and uploading the acquired real-time high-frequency data to the cloud platform if the acquired real-time high-frequency data is detected to be the anomaly high-frequency data; otherwise, the high-frequency data is directly stored to an edge computing platform;
after the cloud platform receives the abnormal high-frequency data of the energy station equipment, the energy station equipment abnormality identification is carried out through a pre-established energy station equipment abnormality identification model, and the abnormal type of the energy station equipment is obtained.
Further, the data acquisition module in the edge computing platform acquires relevant data information of the energy station equipment according to a predefined energy station equipment acquisition object, high frequency data acquisition content, high frequency data acquisition frequency, high frequency data acquisition start-stop conditions and high frequency data acquisition priority, so as to obtain the high frequency data of the energy station equipment, and the method comprises the following steps:
Receiving a preset data acquisition definition file by a data acquisition module in an edge computing platform, wherein the data acquisition definition file comprises an energy station equipment acquisition object, high-frequency data acquisition content, high-frequency data acquisition frequency, high-frequency data acquisition start-stop conditions and high-frequency data acquisition priority;
modeling each energy station equipment acquisition object virtually as an acquisition meta-model, identifying an acquisition meta-model ID, registering in an edge computing platform, detecting whether the acquisition meta-model ID record is owned after the registration is successful, and acquiring data after the registration is confirmed; the acquisition meta model comprises high-frequency data acquisition content, high-frequency data acquisition frequency, high-frequency data acquisition start-stop conditions and high-frequency data acquisition priority related to the acquisition object of the energy station equipment;
according to the high-frequency data acquisition priority, selecting high-priority, medium-priority and low-priority energy station equipment step by step, and according to the high-frequency data acquisition start-stop conditions, when the state of the energy station equipment accords with the acquisition starting conditions, tracking and acquiring the state change of the high-frequency data of the corresponding energy station equipment object according to the high-frequency data acquisition frequency at a preset high-frequency sampling frequency to acquire the high-frequency data of the energy station equipment related to the high-frequency data acquisition content, and when the state of the energy station equipment accords with the acquisition stopping conditions, stopping the high-frequency data acquisition of the corresponding energy station equipment object to generate the high-frequency data of the energy station equipment;
Wherein, the setting of the high frequency data acquisition priority comprises: acquiring an experience value variable interval of an expert for operating the energy station equipment, wherein the experience value variable interval is [0,1],0 is the energy station equipment object with the lowest priority, and 1 is the energy station equipment object with the highest priority; setting the priority of collected data according to the variable interval of the expert experience value of the energy station equipment operation, wherein the priority comprises a high priority, a medium priority and a low priority;
the high-frequency data acquisition start-stop conditions comprise acquisition start marks, acquisition end marks, acquisition suspension marks and acquisition recovery marks.
Further, the edge computing platform further comprises a high-precision time service module, a mobile communication module, a WIFI communication module, a sensor network communication module, a data processing and control module, an external interface module, an equipment position module and a storage module; the high-precision time service module is used for providing a high-precision time synchronization function for the energy station equipment; the mobile communication module is used for supporting communication with a 5G mobile network; the WIFI communication module is used for providing access capability of the wireless local area network; the sensor network communication module is used for providing communication with a high-speed sensor network and equipment; the data processing and control module is used for providing high-performance processing capacity and carrying out data fusion mining processing; the external interface module is used for providing interfaces for connecting various sensors; the equipment position module is used for providing equipment position information through a GPS or Beidou device; the storage module is used for providing a specific storage structure and storing high-frequency data.
Further, after cluster learning is performed on the historical high-frequency data of the energy station equipment through the cloud platform, the high-frequency data is marked with normal and abnormal data labels, and a high-frequency data abnormality detection model is built after training the historical high-frequency data with the labels, and the method comprises the following steps:
acquiring equipment high-frequency data of energy station equipment in a preset historical time period through a cloud platform, and preprocessing;
calculating the distance from the high-frequency data of each energy station device to the initial center by adopting a clustering algorithm, dividing the data into clusters with the minimum distance according to the distance, and iterating;
defining the cluster with the smallest data volume as an abnormal data cluster according to the clustering result, and respectively labeling the normal high-frequency data cluster and the abnormal high-frequency data cluster;
setting parameters of a neural network, and establishing the neural network;
and inputting the clustered high-frequency data of the energy station equipment with the label into a neural network for training, and then establishing a high-frequency data anomaly detection model.
Further, the clustering algorithm adopts a K-means clustering algorithm;
the distance from the high frequency data of each energy station device to the initial center is calculated as follows:
the high-frequency data set of the energy station equipment is defined as X m×n ＝[X 1,n ,X 2,n ,…,X m,n ]The high-frequency data acquisition system consists of m energy station equipment sensors which acquire high-frequency data in n moments; x is X i,n The method comprises the steps that data acquired by an ith energy station equipment sensor in n moments are acquired; c (C) s,n The method comprises the steps that a clustering initial center of high-frequency data of energy station equipment is used;
the normal high frequency data cluster and the abnormal high frequency data cluster are respectively marked with labels, and the labels are expressed as follows:
X m×n ＝[X 1,n ,X 2,n ,…,X m,n ，y i ]＝[(X 1,n ,y i ),(X 1,n ,y i ),…,(X m,n ,y i )]。
further, the neural network adopts an improved BP neural network, a BAS longicorn search algorithm is introduced in the global search process of the FPA flower pollination algorithm, a variation difference strategy is adopted in the local search process to form a BAS-FPA optimization algorithm, and then the BAS-FPA optimization algorithm is utilized to optimize the initial weight and threshold parameters of the BP neural network to form a BAS-FPA-BP network model;
wherein, in the BAS-FPA optimization algorithm, each pollen particle is a solution in a solution space, and represents a probability combination of a weight and a threshold value of a group of BP neural networks;
and (3) carrying out coding processing on the weight and the threshold of the initial BP neural network, establishing a mapping relation among the weight and the threshold of the BP neural network and the BAS-FPA pollen particles in a vector coding mode, and obtaining the optimal weight and the threshold by optimizing and solving to obtain the optimal BP neural network.
Further, after the abnormal high-frequency data of the energy station equipment is received through the cloud platform, the abnormal identification of the energy station equipment is performed through a pre-established abnormal identification model of the energy station equipment, so as to obtain the abnormal type of the energy station equipment, which comprises the following steps:
obtaining an abnormal data sample based on a historical energy station equipment abnormal type database in the cloud platform; the abnormal data sample comprises an abnormal type of energy station equipment corresponding to abnormal high frequency data, wherein the abnormal type at least comprises damage caused by impact of foreign matters in a water pump, surge, loosening of parts, insufficient flow, insufficient air quantity of a fan, overlarge vibration, overheat operation, abrasion of parts, locked rotation of a compressor, low exhaust pressure, overcurrent, part failure, high ambient temperature, leakage and damage of a pipeline;
establishing an improved generation countermeasure network model: taking the generated countermeasure network GAN as a model main body, wherein the model main body comprises a generator network G and a discriminator network D; the generator network G comprises a decoder and an encoder which are sequentially connected, and an independent circulating neural network module is adopted; the discriminator network D comprises a linear layer, an activation function layer, a linear layer and an activation function layer which are sequentially connected; introducing a gradient penalty into the generation of the antagonism network GAN, the loss function of the generator network G being defined as: f w (x) For independent cyclic neural networks, p g To generate data, x is a high frequency data sample; the loss function of the arbiter network D is defined as:
lambda is a gradient penalty function;as a function of the gradient; I.I 2 Is a binary norm of the matrix;
after inputting the abnormal data sample into the improved generation countermeasure network model for training, establishing an energy station equipment abnormal identification model;
when the cloud platform receives the abnormal high-frequency data of the energy station equipment uploaded by the edge computing platform, the energy station equipment abnormality identification model is used for carrying out abnormality identification on the energy station equipment, and the abnormal type of the energy station equipment is output.
Further, the improved generation of the countermeasure network model further includes: introducing a double time scale update rule balances the update speeds of the generator network G and the arbiter network D, expressed as:l d 、l g learning rates of the arbiter network D and the generator network G, respectively; n is the iteration number; h is a n (d)、h n (g) Respectively isGradients of the arbiter network D and the generator network G; beta, & gt>Update coefficients for the arbiter network D and the generator network G, respectively.
Further, after cluster learning is performed on the historical high frequency data of the energy station device through the cloud platform, the high frequency data is marked with normal and abnormal data labels, and the method can also be adopted and comprises the following steps:
Acquiring equipment high-frequency data of energy station equipment in a preset historical time period through a cloud platform, and preprocessing;
extracting the characteristics of the high-frequency data of each energy station device to obtain high-frequency characteristic value data;
clustering high-frequency characteristic value data of a normal state and an abnormal state of the pre-configured energy station equipment by adopting a clustering algorithm to obtain a normal state clustering center and an abnormal state clustering center;
calculating the distance between the high-frequency characteristic value data and the normal state clustering center and the abnormal state clustering center respectively, wherein the distance is expressed as:
(x 3 ,y 3 ) Is a high-frequency eigenvalue data point; d (D) A High frequency eigenvalue data points (x 3 ,y 3 ) Cluster center a (x 1 ,y 1 ) Is a distance of (2); d (D) B High frequency eigenvalue data points (x 3 ,y 3 ) Clustering with abnormal state center B (x 2 ,y 2 ) Is a distance of (2);
if D A ＜D B Indicating that no abnormality has occurred; if D A ＞D B Indicating occurrence ofAbnormality;
and defining the high-frequency data corresponding to the occurrence of the abnormality as an abnormal data cluster according to the clustering result, and respectively labeling the normal high-frequency data cluster and the abnormal high-frequency data cluster.
Further, the edge computing platform and the cloud platform are in secure communication through an authentication and encryption algorithm: the cloud platform generates a unique identity for each edge computing platform and encrypts and transmits the unique identity to the corresponding edge computing platform; the edge computing platform obtains the identity through decryption, attaches the identity to the communication data, and encrypts and transmits the communication data to the cloud platform; after receiving the encrypted data packet, the cloud platform verifies whether the identity is legal through decryption, and after the identity is legal, the data is subjected to subsequent processing.
The beneficial effects of the invention are as follows:
according to the invention, vibration data, sound data and pressure wave data information of the energy station equipment are obtained by installing a vibration sensor, a sound sensor and a pressure wave sensor at the energy station equipment; acquiring related energy station equipment data information according to a predefined energy station equipment acquisition object, high-frequency data acquisition content, high-frequency data acquisition frequency, high-frequency data acquisition start-stop conditions and high-frequency data acquisition priority by a data acquisition module in an edge computing platform to obtain high-frequency data of the energy station equipment; the high-frequency data of the energy station equipment at least comprises sound data, vibration data and pressure wave data of a water pump, a fan, a pipeline and a compressor in the energy station; after cluster learning is carried out on the historical high-frequency data of the energy station equipment through the cloud platform, marking normal and abnormal data labels on the high-frequency data, training the historical high-frequency data with the labels, establishing a high-frequency data abnormal detection model, and transmitting the high-frequency data abnormal detection model to the edge computing platform; detecting the acquired real-time high-frequency data of the energy station equipment through a high-frequency data anomaly detection model in the edge computing platform, and uploading the acquired real-time high-frequency data to the cloud platform if the acquired real-time high-frequency data is detected to be the anomaly high-frequency data; otherwise, the high-frequency data is directly stored to an edge computing platform; after the cloud platform receives the abnormal high-frequency data of the energy station equipment, the abnormal identification of the energy station equipment is carried out through a pre-established abnormal identification model of the energy station equipment, and the abnormal type of the energy station equipment is obtained; on one hand, a high-frequency acquisition data service platform is constructed based on edge calculation to acquire, store, upload, calculate and analyze sensing data, so that the requirement of high-density sensors of energy stations for long-time high-frequency acquisition is met; on the other hand, the main task of the edge computing platform is to collect high-frequency data of the energy station equipment, detect which high-frequency data of the energy station equipment are abnormal, upload the abnormal high-frequency data to the cloud platform, train an abnormality detection model and identify the abnormality type in the cloud platform. The method aims at reducing the flow rate cost of the internet of things transmission, if the high-frequency data of the energy station equipment is uploaded for hundreds of G for one month, if the high-frequency data is only selected for uploading through anomaly detection, the method has practical economic significance on computing resources, storage resources and flow rate cost; therefore, the edge computing platform and the cloud platform cooperatively operate, so that the data acquisition and local data detection processing capacity of the edge computing platform can be fully utilized, the big data processing capacity, model training capacity and data sharing capacity of the cloud platform can be utilized, and finally, the acquisition, anomaly detection and anomaly state identification of high-frequency data are realized.
Additional features and advantages will be set forth in the description which follows, and in part will be apparent from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and drawings.
In order to make the above objects, features and advantages of the present invention more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a cloud edge cooperation-based energy station high frequency data acquisition and abnormal state identification method;
fig. 2 is a schematic block diagram of the energy station high frequency data acquisition and abnormal state identification principle based on cloud edge cooperation.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1
Fig. 1 is a flow chart of a cloud edge cooperation-based method for collecting high-frequency data and identifying abnormal states of an energy station.
Fig. 2 is a schematic block diagram of the energy station high frequency data acquisition and abnormal state identification principle based on cloud edge cooperation.
As shown in fig. 1 and 2, embodiment 1 provides a cloud edge collaboration-based energy station high frequency data acquisition and abnormal state identification method, which includes: a high frequency data acquisition stage and an abnormal state identification stage;
the high frequency data acquisition phase comprises the following steps:
acquiring vibration data, sound data and pressure wave data information of the energy station equipment by installing a vibration sensor, a sound sensor and a pressure wave sensor at the energy station equipment;
Acquiring related energy station equipment data information according to a predefined energy station equipment acquisition object, high-frequency data acquisition content, high-frequency data acquisition frequency, high-frequency data acquisition start-stop conditions and high-frequency data acquisition priority by a data acquisition module in an edge computing platform to obtain high-frequency data of the energy station equipment; the high-frequency data of the energy station equipment at least comprises sound data, vibration data and pressure wave data of a water pump, a fan, a pipeline and a compressor in the energy station;
the abnormal state identification phase comprises the following steps:
after cluster learning is carried out on the historical high-frequency data of the energy station equipment through the cloud platform, marking normal and abnormal data labels on the high-frequency data, training the historical high-frequency data with the labels, establishing a high-frequency data abnormal detection model, and transmitting the high-frequency data abnormal detection model to the edge computing platform;
detecting the acquired real-time high-frequency data of the energy station equipment through a high-frequency data anomaly detection model in the edge computing platform, and uploading the acquired real-time high-frequency data to the cloud platform if the acquired real-time high-frequency data is detected to be the anomaly high-frequency data; otherwise, the high-frequency data is directly stored to an edge computing platform;
after the cloud platform receives the abnormal high-frequency data of the energy station equipment, the energy station equipment abnormality identification is carried out through a pre-established energy station equipment abnormality identification model, and the abnormal type of the energy station equipment is obtained.
It should be noted that, the number of the energy stations is a plurality, the number of the edge computing platforms is at least one, the edge computing platforms and the energy stations belong to a one-to-many or one-to-one relationship, according to the distance between the energy stations, one edge computing platform can be set to correspond to a plurality of energy stations, or one edge computing platform corresponds to one energy station, the nearby collection and the anomaly detection of the high frequency data can be realized through the corresponding edge computing platform, if the detection is the anomaly high frequency data, the operation state of the equipment in the energy station is indicated to have a certain degree of anomaly, and the method belongs to the local identification of the corresponding energy station by each edge computing platform; here, there is a certain correlation between the abnormally high frequency data and the operating state of the energy station apparatus. And carrying out anomaly identification on the anomaly high frequency data through the cloud platform to obtain the anomaly type of the energy station equipment. The energy station equipment anomaly identification model established by the cloud platform belongs to global identification, and the anomaly high-frequency data in the cloud platform is a big data sample of each energy station equipment.
In this embodiment, the acquiring, by a data acquisition module in the edge computing platform, relevant data information of the energy station device according to a predefined energy station device acquisition object, high frequency data acquisition content, high frequency data acquisition frequency, high frequency data acquisition start-stop conditions and high frequency data acquisition priority, to obtain high frequency data of the energy station device includes:
Receiving a preset data acquisition definition file by a data acquisition module in an edge computing platform, wherein the data acquisition definition file comprises an energy station equipment acquisition object, high-frequency data acquisition content, high-frequency data acquisition frequency, high-frequency data acquisition start-stop conditions and high-frequency data acquisition priority;
modeling each energy station equipment acquisition object virtually as an acquisition meta-model, identifying an acquisition meta-model ID, registering in an edge computing platform, detecting whether the acquisition meta-model ID record is owned after the registration is successful, and acquiring data after the registration is confirmed; the acquisition meta model comprises high-frequency data acquisition content, high-frequency data acquisition frequency, high-frequency data acquisition start-stop conditions and high-frequency data acquisition priority related to the acquisition object of the energy station equipment;
according to the high-frequency data acquisition priority, selecting high-priority, medium-priority and low-priority energy station equipment step by step, and according to the high-frequency data acquisition start-stop conditions, when the state of the energy station equipment accords with the acquisition starting conditions, tracking and acquiring the state change of the high-frequency data of the corresponding energy station equipment object according to the high-frequency data acquisition frequency at a preset high-frequency sampling frequency to acquire the high-frequency data of the energy station equipment related to the high-frequency data acquisition content, and when the state of the energy station equipment accords with the acquisition stopping conditions, stopping the high-frequency data acquisition of the corresponding energy station equipment object to generate the high-frequency data of the energy station equipment;
Wherein, the setting of the high frequency data acquisition priority comprises: acquiring an experience value variable interval of an expert for operating the energy station equipment, wherein the experience value variable interval is [0,1],0 is the energy station equipment object with the lowest priority, and 1 is the energy station equipment object with the highest priority; setting the priority of collected data according to the variable interval of the expert experience value of the energy station equipment operation, wherein the priority comprises a high priority, a medium priority and a low priority;
the high-frequency data acquisition start-stop conditions comprise acquisition start marks, acquisition end marks, acquisition suspension marks and acquisition recovery marks.
In this embodiment, the edge computing platform further includes a high-precision time service module, a mobile communication module, a WIFI communication module, a sensor network communication module, a data processing and control module, an external interface module, an equipment location module, and a storage module; the high-precision time service module is used for providing a high-precision time synchronization function for the energy station equipment; the mobile communication module is used for supporting communication with a 5G mobile network; the WIFI communication module is used for providing access capability of the wireless local area network; the sensor network communication module is used for providing communication with a high-speed sensor network and equipment; the data processing and control module is used for providing high-performance processing capacity and carrying out data fusion mining processing; the external interface module is used for providing interfaces for connecting various sensors; the equipment position module is used for providing equipment position information through a GPS or Beidou device; the storage module is used for providing a specific storage structure and storing high-frequency data.
In this embodiment, after cluster learning is performed on the historical high-frequency data of the energy station device through the cloud platform, the high-frequency data is marked with normal and abnormal data labels, and after training is performed on the historical high-frequency data with the labels, a high-frequency data anomaly detection model is built, including:
acquiring equipment high-frequency data of energy station equipment in a preset historical time period through a cloud platform, and preprocessing;
calculating the distance from the high-frequency data of each energy station device to the initial center by adopting a clustering algorithm, dividing the data into clusters with the minimum distance according to the distance, and iterating;
defining the cluster with the smallest data volume as an abnormal data cluster according to the clustering result, and respectively labeling the normal high-frequency data cluster and the abnormal high-frequency data cluster;
setting parameters of a neural network, and establishing the neural network;
and inputting the clustered high-frequency data of the energy station equipment with the label into a neural network for training, and then establishing a high-frequency data anomaly detection model.
It should be noted that, K initial clustering centers are randomly selected from the high-frequency data of the energy station device, the distances from the initial high-frequency data to the centroid are compared and classified, then the clustering center points are updated by calculating the average distance of the high-frequency data in the cluster, and iterating is performed until the final classification result meets the minimum data distance in the cluster, and the iterating process is terminated when the inter-cluster distance is the maximum. As an unsupervised learning method in machine learning, K-means clustering does not need priori knowledge of data, and is simple to calculate, and the characteristics are meeting the requirements for detecting abnormal high frequency data of energy station equipment. According to the fact that the sensor abnormal high-frequency data is smaller than normal data, when the K-means clustering algorithm is applied to abnormal high-frequency data detection of energy station equipment, one cluster with the smallest data quantity in the classified cluster can be considered as an abnormal cluster, and all high-frequency data in the cluster are abnormal data.
The high-frequency data detection of the energy station equipment is realized by combining K-means clustering with a neural network. The K-means clustering is mainly used for realizing normal and abnormal marking of high-frequency data of energy station equipment, and clustering results are used as a sample data set so as to effectively realize modeling analysis of the neural network. The algorithm firstly classifies high-frequency data of the energy station equipment by using a K-means clustering algorithm in unsupervised learning, performs labeling treatment on different categories, trains a neural network by using the high-frequency data of the energy station equipment with labels, obtains an abnormal detection model of the high-frequency data of the energy station equipment, and transmits the abnormal detection model to an edge computing platform, wherein the model does not need clustering treatment again in subsequent detection, and can directly detect the acquired high-frequency data of the energy station equipment.
The main task of the edge computing platform is to detect which energy station equipment high frequency data are abnormal, the abnormal high frequency data are required to be uploaded to the cloud platform, and the training abnormality detection model and the abnormality type identification are on the cloud platform. The purpose of this is to reduce the traffic costs of the internet of things transmission. If the high-frequency data of the energy station equipment is uploaded for several hundred G for one month, if the high-frequency data is uploaded by only selecting abnormal high-frequency data through some detection, the method has practical economic significance on computing resources, storage resources and flow rate fees.
In the embodiment, the clustering algorithm is a K-means clustering algorithm;
the distance from the high frequency data of each energy station device to the initial center is calculated as follows:
the high-frequency data set of the energy station equipment is defined as X m×n ＝[X 1,n ,X 2,n ,…,X m,n ]The high-frequency data acquisition system consists of m energy station equipment sensors which acquire high-frequency data in n moments; x is X i,n The method comprises the steps that data acquired by an ith energy station equipment sensor in n moments are acquired; c (C) s,n The method comprises the steps that a clustering initial center of high-frequency data of energy station equipment is used;
the normal high frequency data cluster and the abnormal high frequency data cluster are respectively marked with labels, and the labels are expressed as follows:
X m×n ＝[X 1,n ,X 2,n ,…,X m,n ，y i ]＝[(X 1,n ,y i ),(X 1,n ,y i ),…,(X m,n ,y i )]。
in the embodiment, the neural network adopts an improved BP neural network, a BAS longhorn whisker search algorithm is introduced in the global search process of an FPA flower pollination algorithm, a variation difference strategy is adopted in the local search process to form a BAS-FPA optimization algorithm, and then the BAS-FPA optimization algorithm is utilized to optimize the initial weight and threshold parameters of the BP neural network to form a BAS-FPA-BP network model;
wherein, in the BAS-FPA optimization algorithm, each pollen particle is a solution in a solution space, and represents a probability combination of a weight and a threshold value of a group of BP neural networks;
and (3) carrying out coding processing on the weight and the threshold of the initial BP neural network, establishing a mapping relation among the weight and the threshold of the BP neural network and the BAS-FPA pollen particles in a vector coding mode, and obtaining the optimal weight and the threshold by optimizing and solving to obtain the optimal BP neural network.
It should be noted that the FPA flower pollination algorithm is a swarm intelligent optimization algorithm inspired by simulating the pollination process of flowering plants in nature. To improve efficiency, the algorithm simplifies the flower pollination process, assuming that only one flower per plant is set up and that only one pollen particle is produced per flower. One flower or one pollen particle may be assigned to one candidate solution in the solution space, and each flower may be propagated for offspring by cross pollination with a transition probability p, or self pollination with probabilities 1-p. Cross pollination can be regarded as the global probing of an algorithm, and self pollination can be regarded as the local searching process of an algorithm, wherein the former refers to the transfer of pollen from one flower to a flower of another plant, and the latter refers to the transfer of pollen from one flower to a different flower of the same plant.
The principle of the FPA algorithm is based on the above process, namely that more age-appropriate flowers in a population survive through pollen transmission, and the main principles include:
cross pollination belongs to a global search process, and global search is performed based on Levy distribution;
self-pollination belongs to a local search process;
the probability of successful pollination of pollen is positively correlated with the similarity of two flowers;
The mutual switching of the global search and the local search is controlled by the transition probability p.
The BAS algorithm does not need to know the specific form of the function, only needs one individual, and compared with other intelligent group algorithms, the calculation amount is greatly reduced. The longicorn is one of beetles, most longicorn has very long antenna for detecting food smell, its bionic principle inspires in the longicorn foraging process and judges the position of food and remove according to the intensity of food smell, every longicorn has two antenna to stretch about two directions, the moving path of longicorn next step always faces the direction that smell intensity is bigger, can regard food smell as a function, the purpose of longicorn is to find the biggest point of global smell intensity, namely the function value is biggest.
The flower pollination algorithm based on the longhorn beetle whisker search improves the convergence rate of the FPA algorithm by introducing the BAS algorithm into the global search process of the FPA algorithm. And improving the ability of the FPA algorithm to jump out of the local optimum by adopting a variation difference strategy in the local search process, thereby finally forming the BAS-FPA algorithm. In the global searching process after improvement, after pollen particles are updated according to the FPA algorithm, each pollen particle can be regarded as a longhorn beetle, a new solution updated according to the FPA algorithm is calculated and compared with a solution before update, if the updated solution is more optimal, the update is carried out, and the solution is reserved, otherwise, the update is abandoned.
In this embodiment, after receiving the abnormal high-frequency data of the energy station device through the cloud platform, performing abnormality identification on the energy station device through a pre-established abnormality identification model of the energy station device, to obtain an abnormality type of the energy station device, including:
obtaining an abnormal data sample based on a historical energy station equipment abnormal type database in the cloud platform; the abnormal data sample comprises an abnormal type of energy station equipment corresponding to abnormal high frequency data, wherein the abnormal type at least comprises damage caused by impact of foreign matters in a water pump, surge, loosening of parts, insufficient flow, insufficient air quantity of a fan, overlarge vibration, overheat operation, abrasion of parts, locked rotation of a compressor, low exhaust pressure, overcurrent, part failure, high ambient temperature, leakage and damage of a pipeline;
establishing an improved generation countermeasure network model: taking the generated countermeasure network GAN as a model main body, wherein the model main body comprises a generator network G and a discriminator network D; the generator network G comprises a decoder and an encoder which are sequentially connected, and an independent circulating neural network module is adopted; the discriminator network D comprises a linear layer, an activation function layer, a linear layer and an activation function layer which are sequentially connected; introducing a gradient penalty into the generation of the antagonism network GAN, the loss function of the generator network G being defined as ：f w (x) For independent cyclic neural networks, p g To generate data, x is a high frequency data sample; the loss function of the arbiter network D is defined as:
lambda is a gradient penalty function;as a function of the gradient; I.I 2 Is a binary norm of the matrix;
after inputting the abnormal data sample into the improved generation countermeasure network model for training, establishing an energy station equipment abnormal identification model;
when the cloud platform receives the abnormal high-frequency data of the energy station equipment uploaded by the edge computing platform, the energy station equipment abnormality identification model is used for carrying out abnormality identification on the energy station equipment, and the abnormal type of the energy station equipment is output.
It should be noted that, applying the gradient penalty to the training generation countermeasure network GAN not only improves the network convergence performance and the quality of the generated data samples, but also makes the network training more stable. And performing unsupervised anomaly identification by adopting a generated countermeasure network structure. The generation network part adopts an independent circulating neural network module, so that the depth of the network structure can be effectively improved, and the problems of gradient disappearance and gradient explosion can be avoided by using an unsaturated function as an activation function. Therefore, the improved generation countermeasure network model solves the problems that the existing training based on the generation countermeasure network is not easy to converge, gradient explosion and the quality of the generated data sample is poor.
In this embodiment, the generating of the improved countermeasure network model further includes: introducing a double time scale update rule balances the update speeds of the generator network G and the arbiter network D, expressed as:l d 、l g learning rates of the arbiter network D and the generator network G, respectively; n is the iteration number; h is a n (d)、h n (g) Gradients of the arbiter network D and the generator network G, respectively; beta, & gt>Update coefficients for the arbiter network D and the generator network G, respectively.
It should be noted that the improved generation countermeasure network model uses the same learning rate in training G and D, but the network itself has the feature that the update speed of D does not keep up with the update speed of G, and the update speed between the two needs to be balanced as much as possible during model training to prevent the network from collapsing. Therefore, a two-time scale update rule is introduced to balance the network update problem, and different learning rates are set for D and G in the WGAN-GP network, so as to accelerate the convergence rate of D, i.e., the learning rate of D is greater than the learning rate of G. The two-time scale updating rule can enable G and D to be updated at the speed of 1:1, and better results are generated under the same time condition, so that the generation of the countermeasure network model is more stable, and the running time is reduced.
In this embodiment, after cluster learning is performed on the historical high-frequency data of the energy station device through the cloud platform, the high-frequency data is marked with normal and abnormal data labels, and a method may also be adopted, including:
acquiring equipment high-frequency data of energy station equipment in a preset historical time period through a cloud platform, and preprocessing;
extracting the characteristics of the high-frequency data of each energy station device to obtain high-frequency characteristic value data;
clustering high-frequency characteristic value data of a normal state and an abnormal state of the pre-configured energy station equipment by adopting a clustering algorithm to obtain a normal state clustering center and an abnormal state clustering center;
calculating the distance between the high-frequency characteristic value data and the normal state clustering center and the abnormal state clustering center respectively, wherein the distance is expressed as:
(x 3 ,y 3 ) Is a high-frequency eigenvalue data point; d (D) A High frequency eigenvalue data points (x 3 ,y 3 ) Cluster center a (x 1 ,y 1 ) Is a distance of (2); d (D) B High frequency eigenvalue data points (x 3 ,y 3 ) Clustering with abnormal state center B (x 2 ,y 2 ) Is a distance of (2);
if D A ＜D B Indicating that no abnormality has occurred; if D A ＞D B Indicating that an abnormality has occurred;
and defining the high-frequency data corresponding to the occurrence of the abnormality as an abnormal data cluster according to the clustering result, and respectively labeling the normal high-frequency data cluster and the abnormal high-frequency data cluster.
In this embodiment, secure communication is performed between the edge computing platform and the cloud platform through an authentication and encryption/decryption algorithm: the cloud platform generates a unique identity for each edge computing platform and encrypts and transmits the unique identity to the corresponding edge computing platform; the edge computing platform obtains the identity through decryption, attaches the identity to the communication data, and encrypts and transmits the communication data to the cloud platform; after receiving the encrypted data packet, the cloud platform verifies whether the identity is legal through decryption, and after the identity is legal, the data is subjected to subsequent processing.
In the several embodiments provided in the present application, it should be understood that the disclosed system and method may be implemented in other manners as well. The system embodiments described above are merely illustrative, for example, of the flowcharts and block diagrams in the figures that illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
In addition, functional modules in the embodiments of the present invention may be integrated together to form a single part, or each module may exist alone, or two or more modules may be integrated to form a single part. The functions, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored on a computer readable storage medium. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method of the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
With the above-described preferred embodiments according to the present invention as an illustration, the above-described descriptions can be used by persons skilled in the relevant art to make various changes and modifications without departing from the scope of the technical idea of the present invention. The technical scope of the present invention is not limited to the description, but must be determined according to the scope of claims.

Claims (10)

1. The cloud edge cooperation-based energy station high-frequency data acquisition and abnormal state identification method is characterized by comprising the following steps of: a high frequency data acquisition stage and an abnormal state identification stage;
the high frequency data acquisition phase comprises the following steps:
acquiring vibration data, sound data and pressure wave data information of the energy station equipment by installing a vibration sensor, a sound sensor and a pressure wave sensor at the energy station equipment;
acquiring related energy station equipment data information according to a predefined energy station equipment acquisition object, high-frequency data acquisition content, high-frequency data acquisition frequency, high-frequency data acquisition start-stop conditions and high-frequency data acquisition priority by a data acquisition module in an edge computing platform to obtain high-frequency data of the energy station equipment; the high-frequency data of the energy station equipment at least comprises sound data, vibration data and pressure wave data of a water pump, a fan, a pipeline and a compressor in the energy station;
the abnormal state identification phase comprises the following steps:
after cluster learning is carried out on the historical high-frequency data of the energy station equipment through the cloud platform, marking normal and abnormal data labels on the high-frequency data, training the historical high-frequency data with the labels, establishing a high-frequency data abnormal detection model, and transmitting the high-frequency data abnormal detection model to the edge computing platform;
Detecting the acquired real-time high-frequency data of the energy station equipment through a high-frequency data anomaly detection model in the edge computing platform, and uploading the acquired real-time high-frequency data to the cloud platform if the acquired real-time high-frequency data is detected to be the anomaly high-frequency data; otherwise, the high-frequency data is directly stored to an edge computing platform;
after the cloud platform receives the abnormal high-frequency data of the energy station equipment, the energy station equipment abnormality identification is carried out through a pre-established energy station equipment abnormality identification model, and the abnormal type of the energy station equipment is obtained.
2. The method for collecting high-frequency data and identifying abnormal states of an energy station according to claim 1, wherein the collecting, by a data collecting module in an edge computing platform, relevant high-frequency data of the energy station according to a predefined object for collecting the energy station, high-frequency data collecting content, high-frequency data collecting frequency, high-frequency data collecting start-stop conditions and high-frequency data collecting priority, to obtain the high-frequency data of the energy station, comprises:
receiving a preset data acquisition definition file by a data acquisition module in an edge computing platform, wherein the data acquisition definition file comprises an energy station equipment acquisition object, high-frequency data acquisition content, high-frequency data acquisition frequency, high-frequency data acquisition start-stop conditions and high-frequency data acquisition priority;
Modeling each energy station equipment acquisition object virtually as an acquisition meta-model, identifying an acquisition meta-model ID, registering in an edge computing platform, detecting whether the acquisition meta-model ID record is owned after the registration is successful, and acquiring data after the registration is confirmed; the acquisition meta model comprises high-frequency data acquisition content, high-frequency data acquisition frequency, high-frequency data acquisition start-stop conditions and high-frequency data acquisition priority related to the acquisition object of the energy station equipment;
according to the high-frequency data acquisition priority, selecting high-priority, medium-priority and low-priority energy station equipment step by step, and according to the high-frequency data acquisition start-stop conditions, when the state of the energy station equipment accords with the acquisition starting conditions, tracking and acquiring the state change of the high-frequency data of the corresponding energy station equipment object according to the high-frequency data acquisition frequency at a preset high-frequency sampling frequency to acquire the high-frequency data of the energy station equipment related to the high-frequency data acquisition content, and when the state of the energy station equipment accords with the acquisition stopping conditions, stopping the high-frequency data acquisition of the corresponding energy station equipment object to generate the high-frequency data of the energy station equipment;
wherein, the setting of the high frequency data acquisition priority comprises: acquiring an experience value variable interval of an expert for operating the energy station equipment, wherein the experience value variable interval is [0,1],0 is the energy station equipment object with the lowest priority, and 1 is the energy station equipment object with the highest priority; setting the priority of collected data according to the variable interval of the expert experience value of the energy station equipment operation, wherein the priority comprises a high priority, a medium priority and a low priority;
The high-frequency data acquisition start-stop conditions comprise acquisition start marks, acquisition end marks, acquisition suspension marks and acquisition recovery marks.
3. The method for collecting high-frequency data and identifying abnormal states of an energy station according to claim 1, wherein the edge computing platform further comprises a high-precision time service module, a mobile communication module, a WIFI communication module, a sensor network communication module, a data processing and control module, an external interface module, a device position module and a storage module; the high-precision time service module is used for providing a high-precision time synchronization function for the energy station equipment; the mobile communication module is used for supporting communication with a 5G mobile network; the WIFI communication module is used for providing access capability of the wireless local area network; the sensor network communication module is used for providing communication with a high-speed sensor network and equipment; the data processing and control module is used for providing high-performance processing capacity and carrying out data fusion mining processing; the external interface module is used for providing interfaces for connecting various sensors; the equipment position module is used for providing equipment position information through a GPS or Beidou device; the storage module is used for providing a specific storage structure and storing high-frequency data.
4. The method for collecting high-frequency data and identifying abnormal states of an energy station according to claim 1, wherein after cluster learning is performed on historical high-frequency data of the energy station device through a cloud platform, the high-frequency data is labeled with normal and abnormal data, and after training is performed on the labeled historical high-frequency data, a high-frequency data abnormal detection model is built, and the method comprises the following steps:
acquiring equipment high-frequency data of energy station equipment in a preset historical time period through a cloud platform, and preprocessing;
calculating the distance from the high-frequency data of each energy station device to the initial center by adopting a clustering algorithm, dividing the data into clusters with the minimum distance according to the distance, and iterating;
defining the cluster with the smallest data volume as an abnormal data cluster according to the clustering result, and respectively labeling the normal high-frequency data cluster and the abnormal high-frequency data cluster;
setting parameters of a neural network, and establishing the neural network;
and inputting the clustered high-frequency data of the energy station equipment with the label into a neural network for training, and then establishing a high-frequency data anomaly detection model.
5. The method for collecting high frequency data and identifying abnormal states of an energy station according to claim 4, wherein the clustering algorithm is a K-means clustering algorithm;
The distance from the high frequency data of each energy station device to the initial center is calculated as follows:
the high-frequency data set of the energy station equipment is defined as X m×n ＝[X 1,n ,X 2,n ,…,X m,n ]The high-frequency data acquisition system consists of m energy station equipment sensors which acquire high-frequency data in n moments; x is X i,n The method comprises the steps that data acquired by an ith energy station equipment sensor in n moments are acquired; c (C) s,n The method comprises the steps that a clustering initial center of high-frequency data of energy station equipment is used;
the normal high frequency data cluster and the abnormal high frequency data cluster are respectively marked with labels, and the labels are expressed as follows:
X m×n ＝[X 1,n ,X 2,n ,…,X m,n ，y i ]＝[(X 1,n ,y i ),(X 1,n ,y i ),…,(X m,n ,y i )]。
6. the method for collecting high frequency data and identifying abnormal states of an energy station according to claim 4, wherein the neural network adopts an improved BP neural network, a BAS longicorn search algorithm is introduced in the global search process of an FPA flower pollination algorithm, a BAS-FPA optimization algorithm is formed after a variation difference strategy is adopted in the local search process, and initial weight and threshold parameters of the BP neural network are optimized by the BAS-FPA optimization algorithm, so that a BAS-FPA-BP network model is formed;
wherein, in the BAS-FPA optimization algorithm, each pollen particle is a solution in a solution space, and represents a probability combination of a weight and a threshold value of a group of BP neural networks;
and (3) carrying out coding processing on the weight and the threshold of the initial BP neural network, establishing a mapping relation among the weight and the threshold of the BP neural network and the BAS-FPA pollen particles in a vector coding mode, and obtaining the optimal weight and the threshold by optimizing and solving to obtain the optimal BP neural network.
7. The method for collecting high-frequency data and identifying abnormal states of energy station according to claim 1, wherein after receiving the abnormal high-frequency data of the energy station equipment through the cloud platform, performing abnormality identification of the energy station equipment through a pre-established abnormality identification model of the energy station equipment to obtain the abnormality type of the energy station equipment, comprising:
obtaining an abnormal data sample based on a historical energy station equipment abnormal type database in the cloud platform; the abnormal data sample comprises an abnormal type of energy station equipment corresponding to abnormal high frequency data, wherein the abnormal type at least comprises damage caused by impact of foreign matters in a water pump, surge, loosening of parts, insufficient flow, insufficient air quantity of a fan, overlarge vibration, overheat operation, abrasion of parts, locked rotation of a compressor, low exhaust pressure, overcurrent, part failure, high ambient temperature, leakage and damage of a pipeline;
establishing an improved generation countermeasure network model: the method takes the generation of the countermeasure network GAN as a model main body and comprises the generation of a network G and judgmentA network D of other devices; the generator network G comprises a decoder and an encoder which are sequentially connected, and an independent circulating neural network module is adopted; the discriminator network D comprises a linear layer, an activation function layer, a linear layer and an activation function layer which are sequentially connected; introducing a gradient penalty into the generation of the antagonism network GAN, the loss function of the generator network G being defined as: f w (x) For independent cyclic neural networks, p g To generate data, x is a high frequency data sample; the loss function of the arbiter network D is defined as:
lambda is a gradient penalty function;as a function of the gradient; I.I 2 Is a binary norm of the matrix;
after inputting the abnormal data sample into the improved generation countermeasure network model for training, establishing an energy station equipment abnormal identification model;
when the cloud platform receives the abnormal high-frequency data of the energy station equipment uploaded by the edge computing platform, the energy station equipment abnormality identification model is used for carrying out abnormality identification on the energy station equipment, and the abnormal type of the energy station equipment is output.
8. The energy station high frequency data collection and abnormal state identification method of claim 7, wherein said improved generation of an countermeasure network model further comprises: introducing a double time scale update rule balances the update speeds of the generator network G and the arbiter network D, expressed as:l d 、l g respectively a discriminator network D and a generator networkThe learning rate of the complex G; n is the iteration number; h is a n (d)、h n (g) Gradients of the arbiter network D and the generator network G, respectively; beta, & gt>Update coefficients for the arbiter network D and the generator network G, respectively.
9. The method for collecting high-frequency data and identifying abnormal states of an energy station according to claim 4, wherein after cluster learning is performed on historical high-frequency data of the energy station device by using a cloud platform, the high-frequency data is marked with normal and abnormal data, and the method further comprises:
Acquiring equipment high-frequency data of energy station equipment in a preset historical time period through a cloud platform, and preprocessing;
extracting the characteristics of the high-frequency data of each energy station device to obtain high-frequency characteristic value data;
clustering high-frequency characteristic value data of a normal state and an abnormal state of the pre-configured energy station equipment by adopting a clustering algorithm to obtain a normal state clustering center and an abnormal state clustering center;
calculating the distance between the high-frequency characteristic value data and the normal state clustering center and the abnormal state clustering center respectively, wherein the distance is expressed as:
(x 3 ,y 3 ) Is a high-frequency eigenvalue data point; d (D) A High frequency eigenvalue data points (x 3 ,y 3 ) Cluster center a (x 1 ,y 1 ) Is a distance of (2); d (D) B Gao Pinte characterised by the current high frequency dataCharacterization value data points (x) 3 ,y 3 ) Clustering with abnormal state center B (x 2 ,y 2 ) Is a distance of (2);
if D A ＜D B Indicating that no abnormality has occurred; if D A ＞D B Indicating that an abnormality has occurred;
and defining the high-frequency data corresponding to the occurrence of the abnormality as an abnormal data cluster according to the clustering result, and respectively labeling the normal high-frequency data cluster and the abnormal high-frequency data cluster.
10. The method for collecting high-frequency data and identifying abnormal states of an energy station according to claim 1, wherein the edge computing platform and the cloud platform are in secure communication through an identity verification and encryption and decryption algorithm: the cloud platform generates a unique identity for each edge computing platform and encrypts and transmits the unique identity to the corresponding edge computing platform; the edge computing platform obtains the identity through decryption, attaches the identity to the communication data, and encrypts and transmits the communication data to the cloud platform; after receiving the encrypted data packet, the cloud platform verifies whether the identity is legal through decryption, and after the identity is legal, the data is subjected to subsequent processing.
CN202310687531.1A 2023-06-09 2023-06-09 Cloud edge cooperation-based energy station high-frequency data acquisition and abnormal state identification method Pending CN116702038A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310687531.1A CN116702038A (en) 2023-06-09 2023-06-09 Cloud edge cooperation-based energy station high-frequency data acquisition and abnormal state identification method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310687531.1A CN116702038A (en) 2023-06-09 2023-06-09 Cloud edge cooperation-based energy station high-frequency data acquisition and abnormal state identification method

Publications (1)

Publication Number Publication Date
CN116702038A true CN116702038A (en) 2023-09-05

Family

ID=87833522

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310687531.1A Pending CN116702038A (en) 2023-06-09 2023-06-09 Cloud edge cooperation-based energy station high-frequency data acquisition and abnormal state identification method

Country Status (1)

Country Link
CN (1) CN116702038A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117914003A (en) * 2024-03-19 2024-04-19 沈阳智帮电气设备有限公司 Intelligent monitoring auxiliary method and system for box-type transformer based on cloud edge cooperation

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117914003A (en) * 2024-03-19 2024-04-19 沈阳智帮电气设备有限公司 Intelligent monitoring auxiliary method and system for box-type transformer based on cloud edge cooperation
CN117914003B (en) * 2024-03-19 2024-05-24 沈阳智帮电气设备有限公司 Intelligent monitoring auxiliary method and system for box-type transformer based on cloud edge cooperation

Similar Documents

Publication Publication Date Title
US20240275805A1 (en) Cognitive neuro-linguistic behavior recognition system for multi-sensor data fusion
Dao et al. Identification failure data for cluster heads aggregation in WSN based on improving classification of SVM
CN106529719B (en) Wind power prediction method based on particle swarm optimization algorithm wind speed fusion
Cao et al. CNN-based intelligent safety surveillance in green IoT applications
CN112990558B (en) Meteorological temperature and illumination prediction method based on deep migration learning
CN112395289B (en) Distributed photovoltaic data layered storage method and system
CN108734359B (en) Wind power prediction data preprocessing method
CN117171686A (en) Method and system for detecting abnormal data of intelligent power grid based on federal learning
CN116862132A (en) Resource scheduling method based on big data
CN116702038A (en) Cloud edge cooperation-based energy station high-frequency data acquisition and abnormal state identification method
CN114021822A (en) Clean energy power generation power prediction method and system
CN118133144A (en) Vehicle fault diagnosis method, device, equipment and medium based on graph neural network
CN117540422A (en) Photovoltaic power generation amount prediction method adopting environment perception under federal learning architecture
CN117097026A (en) Operation method of novel power system operation and maintenance monitoring platform based on source network charge storage
CN116678082A (en) Control method and terminal of central air conditioning system and central air conditioning system
CN110852344A (en) Intelligent substation network fault classification based method
CN111382763A (en) Photovoltaic cluster division method and system
CN112801181B (en) Urban signaling traffic flow user classification and prediction method, storage medium and system
Liu et al. MCT‐TTE: Travel Time Estimation Based on Transformer and Convolution Neural Networks
CN113970799A (en) Bridge meteorological monitoring system, method, equipment and storage medium based on narrow-band communication
CN114139781A (en) Method and system for predicting operation trend of power system
Yuan et al. Hybrid and Spatiotemporal Detection of Cyberattack Network Traffic in Cloud Data Centers
CN105608638B (en) Method and system for judging synchronous state of meter code data of intelligent terminal and electric energy meter
CN115883016A (en) Method and device for enhancing flow data based on federal generation countermeasure network
CN112269778A (en) Equipment fault diagnosis method

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