CN-121998278-A - Unmanned intelligent inspection method and system for water and electricity river basin data center

CN121998278ACN 121998278 ACN121998278 ACN 121998278ACN-121998278-A

Abstract

The invention discloses an unattended intelligent inspection method and system for a hydropower basin data center, and relates to the technical field of intelligent operation and maintenance scheduling, wherein the method comprises the steps of collecting data of all areas of the hydropower basin data center, preprocessing the data, and calculating abnormal values of all areas of the hydropower basin data center based on the preprocessed data; and deducing the minimum detection frequency required by each region by using poisson random distribution based on the abnormal value of each region of the data center of the hydropower basin, carrying out global planning and scheduling on inspection by using NSGA-II algorithm based on the minimum detection frequency, and predicting the fault probability of inspection equipment by using LSTM prediction model in the inspection process. According to the invention, the minimum detection frequency is calculated by introducing a poisson random distribution model, and the global scheduling planning is carried out by combining an NSGA-II multi-objective optimization algorithm, so that the key problems in the aspects of intellectualization of routing inspection scheduling and optimal allocation of equipment resources in the prior art are effectively solved.

Inventors

XIE ZHIQI
TANG XIAOBO
ZENG TIJIAN
DU ZEXIN
LI LIN
LI YUANJUN
ZHU XI
ZHANG YUJI
SU QIAN
LUO YU

Assignees

贵州乌江水电开发有限责任公司

Dates

Publication Date: 20260508
Application Date: 20251128

Claims (10)

1. An unattended intelligent inspection method for a water and electricity river basin data center is characterized by comprising the steps of, Collecting data of each area of the water and electricity river basin data center, preprocessing, and calculating abnormal values of each area of the water and electricity river basin data center based on the preprocessed data; Deducing the minimum detection frequency required by each region by using poisson random distribution based on the abnormal value of each region of the hydroelectric basin data center, and performing global planning and scheduling on patrol inspection by using NSGA-II algorithm based on the minimum detection frequency; In the inspection process, the LSTM prediction model is used for predicting the fault probability of the inspection equipment, meanwhile, the cost-efficiency function is used for calculating the critical point of the deployment scale of the inspection equipment, and the inspection equipment is deployed based on the calculation result.
2. The unattended intelligent inspection method of the hydroelectric basin data center is characterized in that the collection of data of all areas of the hydroelectric basin data center and pretreatment are performed by setting a collection period, collecting water flow and current data of all areas of the hydroelectric basin data center at each moment in the collection period to obtain a water flow and current data time sequence of all areas of the hydroelectric basin data center, normalizing the data time sequence by using a Z-score normalization method, defining a sliding time window for the normalized data time sequence, and calculating standard deviation of data points in the sliding time window ; Setting standard deviation threshold If (if) Will then The data point at the moment is taken as normal data, if Starting to perform abnormality detection, setting M continuous detection moments, and if the standard deviation of the data points is greater than or equal to the set standard deviation threshold value in the continuous detection moments, setting M continuous detection moments The data points in the moment and the continuous detection moment are taken as effective abnormal data points, if the standard deviation of the data points in the continuous detection moment is smaller than the set standard deviation threshold value, the data points are taken as effective abnormal data points The data point at the moment is taken as invalid outlier data point.
3. The unattended intelligent inspection method of the hydropower basin data center is characterized in that the calculating of abnormal values of all areas of the hydropower basin data center based on the pretreatment data means that data points judged to be effective abnormal are placed into an abnormal data point set, and the abnormal values of all areas of the hydropower basin data center are calculated according to the number of the abnormal data points in the abnormal data point set.
4. The unattended intelligent inspection method of the water and electricity river basin data center according to claim 3, wherein the minimum detection frequency required by deducing each area by poisson random distribution based on the abnormal value of each area of the water and electricity river basin data center is calculated based on the abnormal value of each area of the water and electricity river basin data center, and the detection frequency is calculated based on the detection frequency of each area in unit time; modeling the detection behavior of the inspection equipment on each area as poisson distribution according to the times that each area needs to be detected in unit time; Calculating the probability that each region is detected at least once in a unit time based on the poisson distribution function; Setting a probability threshold value to be detected at least once per unit time Based on probability threshold And calculating the lowest detection frequency of each region.
5. The unmanned intelligent inspection method of the hydroelectric basin data center of claim 4, wherein the global planning and dispatching for inspection based on the lowest detection frequency by using NSGA-II algorithm means that an inspection dispatching scheme set is generated under the condition that the lowest detection frequency of each region is met, and an optimized objective function value is calculated for each scheme in the inspection dispatching scheme set, including the total path length, the task balance and the running cost; non-dominant sorting is carried out on all the patrol schemes in the patrol scheduling scheme set to obtain the non-dominant patrol scheduling scheme set, and the sparseness degree of the schemes in the non-dominant patrol scheduling scheme set is calculated ; Setting a sparseness threshold If (if) Putting the scheme into a preferable patrol scheduling scheme set, if No scheme is adopted; and in the preferred patrol scheduling scheme set, generating a new patrol scheduling scheme by adopting intersection and mutation operations, carrying out non-dominant sorting and sparseness screening again, repeatedly executing the operations, and outputting the patrol scheduling scheme with the largest sparseness degree and non-dominant in the preferred patrol scheduling scheme set as an optimal patrol scheduling scheme after the iteration number reaches a set upper limit.
6. The unmanned intelligent patrol method of the hydroelectric basin data center of claim 5, wherein in the patrol process, the use of the LSTM prediction model to predict the fault probability of the patrol equipment refers to outputting a state sequence of the patrol equipment in real time, and the hidden state of the patrol equipment is calculated based on the state sequence of the patrol equipment; Setting a prediction time, and calculating the probability of fault of the inspection equipment within the prediction time based on the hidden state of the inspection equipment ; Setting a failure probability threshold If (if) The inspection equipment is normally used, if And identifying the inspection equipment as high-risk equipment, and not participating in inspection scheduling and inspection maintenance.
7. The unattended intelligent tour inspection method of a hydropower basin data center according to claim 6, wherein the critical point of the deployment scale of the tour inspection equipment is calculated by using a cost-efficiency function, the cost function is calculated according to the operation cost and the equipment cost based on the calculation result, the unit efficiency cost is calculated by using the cost function and the efficiency function according to the number of the deployment of the tour inspection equipment, the unit efficiency cost is derived, the minimum value point with the derivative of zero is solved, and the number of the tour inspection equipment deployed by the extreme point is used as the optimal deployment.
8. An unattended intelligent inspection system of a hydropower basin data center is based on the unattended intelligent inspection method of the hydropower basin data center according to any one of claims 1-7, and is characterized by comprising, The data acquisition module is used for acquiring the data of each region of the hydropower basin data center and carrying out normalization processing; the abnormal value detection module is used for detecting abnormal data in the acquired data and judging whether the abnormal data are effective abnormal data points or ineffective abnormal data points; the detection frequency deducing module is used for calculating the lowest detection frequency of each region by using poisson random distribution according to the abnormal value of each region; The inspection scheduling module is used for globally optimizing the inspection scheduling by using an NSGA-II algorithm and generating an optimal inspection scheduling scheme by taking the lowest detection frequency as a constraint condition; the fault prediction module is used for predicting faults of the inspection equipment by adopting an LSTM model according to the state of the inspection equipment; and the equipment deployment module is used for calculating the optimal deployment quantity of the inspection equipment by using the cost-efficiency function.
9. The computer equipment comprises a memory and a processor, wherein the memory stores a computer program, and the computer program is characterized in that the processor realizes the steps of the unattended intelligent inspection method of the hydropower basin data center according to any one of claims 1-7 when executing the computer program.
10. A computer readable storage medium is provided, and is characterized in that the computer program is executed by a processor to realize the steps of the unattended intelligent inspection method of the hydropower basin data center according to any one of claims 1-7.

Description

Unmanned intelligent inspection method and system for water and electricity river basin data center Technical Field The invention relates to the technical field of intelligent operation and maintenance scheduling, in particular to an unattended intelligent inspection method and system for a hydropower basin data center. Background With the wide application of the Internet of things, big data and artificial intelligence technology, a hydropower basin data center becomes an important infrastructure for basin scheduling, ecological monitoring, safety early warning and comprehensive management. Most of the existing inspection systems carry out inspection path planning based on a preset time period or manual experience, and basic running state detection is realized by arranging a plurality of fixed detection devices or unmanned inspection terminals, but new challenges are faced in detection frequency scheduling, resource allocation and equipment operation and maintenance efficiency along with the expansion of the scale of a data center and the increase of the diversity of the equipment. At present, a common unattended inspection scheme is characterized in that inspection tasks are determined through statistical features or an anomaly detection model based on rules, and task scheduling is performed by combining a heuristic algorithm or a genetic algorithm. Although the method can realize automatic inspection to a certain extent, under a multi-objective optimization scene, balanced scheduling is difficult to realize at a global level. Especially in the multi-node heterogeneous environment of the hydropower basin data center, on the premise of ensuring the stability of the system, the self-adaptive planning of the inspection frequency and the dynamic optimization of the resource allocation are realized, and the method is still an important research direction in the industry. Disclosure of Invention The present invention has been made in view of the above-described problems occurring in the prior art. Therefore, the invention provides an unattended intelligent inspection method for a hydropower basin data center, which solves the problems of insufficient detection frequency self-adaptive planning and inconsistent equipment deployment scale optimization in inspection of the hydropower basin data center. In order to solve the technical problems, the invention provides the following technical scheme: In a first aspect, the invention provides an unattended intelligent inspection method for a hydropower basin data center, which comprises the steps of, Collecting data of each area of the water and electricity river basin data center, preprocessing, and calculating abnormal values of each area of the water and electricity river basin data center based on the preprocessed data; Deducing the minimum detection frequency required by each region by using poisson random distribution based on the abnormal value of each region of the hydroelectric basin data center, and performing global planning and scheduling on patrol inspection by using NSGA-II algorithm based on the minimum detection frequency; In the inspection process, the LSTM prediction model is used for predicting the fault probability of the inspection equipment, meanwhile, the cost-efficiency function is used for calculating the critical point of the deployment scale of the inspection equipment, and the inspection equipment is deployed based on the calculation result. The invention relates to an unattended intelligent inspection method for a hydroelectric basin data center, which comprises the following steps of collecting data of each area of the hydroelectric basin data center and preprocessing, setting a collection period, collecting water flow and current data of each area of the hydroelectric basin data center at each moment in the collection period to obtain a water flow and current data time sequence of each area of the hydroelectric basin data center, normalizing the data time sequence by using a Z-score normalization method, defining a sliding time window aiming at the normalized data time sequence, and calculating standard deviation of data points in the sliding time window; Setting standard deviation thresholdIf (if)Will thenThe data point at the moment is taken as normal data, ifStarting to perform abnormality detection, setting M continuous detection moments, and if the standard deviation of the data points is greater than or equal to the set standard deviation threshold value in the continuous detection moments, setting M continuous detection momentsThe data points in the moment and the continuous detection moment are taken as effective abnormal data points, if the standard deviation of the data points in the continuous detection moment is smaller than the set standard deviation threshold value, the data points are taken as effective abnormal data pointsThe data point at the moment is taken as invalid outlier data point. The method for unattended intellig