CN-121835219-B - Intelligent prediction method and system for optical cable performance of data center

Abstract

The invention provides an intelligent prediction method and system for optical cable performance of a data center, which relate to the technical field of performance analysis of the data center, and comprises the steps of firstly constructing an optical cable microstructure disturbance quantization model, longitudinally dividing the optical cable into a plurality of virtual quantization units, packaging real-time stress strain state parameters and response sensitivity coefficients of basic structural units by each unit, inputting a real-time acquired Brillouin scattering spectrum frequency shift data stream into the model for disturbance source analysis, generating a physical excitation decomposition sequence, driving a virtual quantification unit to perform state evolution iterative operation according to the physical excitation decomposition sequence to obtain a stress strain state evolution track set, constructing a space-time correlation network for optical cable link disturbance propagation according to the stress strain state evolution track set, performing network flow characteristic analysis on the space-time correlation network, extracting an early warning node set, calculating future evolution trend of the early warning node set, and generating serialized optical cable performance abnormality early warning information. The invention can predict the abnormal performance of the optical cable in advance and ensure the stable operation of the optical cable of the data center.

Inventors

• LI DEJIAN
• HE JIN

Assignees

• 四川省嘉万光通信有限公司

Dates

Publication Date

20260512

Application Date

20260313

Claims (10)

1. 1. An intelligent prediction method for optical cable performance of a data center, the method comprising: Constructing an optical cable microstructure disturbance quantization model, wherein the optical cable microstructure disturbance quantization model comprises a plurality of virtual quantization units distributed along the longitudinal direction of an optical cable, each virtual quantization unit corresponds to a basic structural unit in the optical cable, and each virtual quantization unit is internally packaged with a real-time stress strain state parameter of the basic structural unit and a response sensitivity coefficient of the basic structural unit to external disturbance; Inputting the optical cable acquired in real time into the optical cable microstructure disturbance quantization model along the line Brillouin scattering spectrum frequency shift data stream to perform disturbance source analysis processing, and generating a physical excitation decomposition sequence received by each virtual quantization unit at the current moment, wherein each element in the physical excitation decomposition sequence comprises a temperature disturbance component and a strain disturbance component which are obtained by analyzing the positions of the corresponding virtual quantization units; Driving virtual quantization units in the optical cable microstructure disturbance quantization model to perform state evolution iterative operation according to the physical excitation decomposition sequence to obtain a stress strain state evolution track set of each virtual quantization unit on a continuous time axis, wherein the stress strain state evolution track set comprises a stress tensor time-varying curve and a strain tensor time-varying curve of each virtual quantization unit; constructing a space-time correlation network for disturbance propagation of an optical cable link based on stress-strain state differences between adjacent virtual quantification units in the stress-strain state evolution track set, wherein the space-time correlation network takes the virtual quantification units as nodes, the change rate of the stress-strain state differences between adjacent nodes is taken as a weight value of node connecting edges, and the space-time correlation network is used for describing a transmission path and a transmission rate of the disturbance along the longitudinal direction of the optical cable; And carrying out network flow characteristic analysis processing on the space-time correlation network, extracting an early warning node set in which the stress strain state in the network nodes is accumulated to exceed a critical threshold, calculating the evolution trend of the stress strain state of each node in the early warning node set in a future time window, and generating serialized optical cable performance abnormality early warning information containing early warning node position coordinates and a stress strain state predicted value according to the evolution trend.
2. 2. The intelligent prediction method for optical cable performance of data center according to claim 1, wherein the constructing an optical cable microstructure disturbance quantization model comprises: Dividing the internal physical structure of the optical cable of the data center into a continuous basic structure unit sequence along the longitudinal direction, wherein the length of each basic structure unit is matched with the microstructure characteristic scale of the optical cable material, each basic structure unit in the basic structure unit sequence corresponds to one virtual quantization unit, and the total number of the virtual quantization units is obtained by dividing the total length of the optical cable by the length of the basic structure unit; allocating a unique space index identifier for each virtual quantization unit, wherein the space index identifiers are increased gradually according to the direction from the starting end to the tail end of the optical cable, and each space index identifier establishes a mapping relation with the starting point coordinate and the end point coordinate of the actual physical interval of the optical cable covered by the virtual quantization unit to form a corresponding table of the space index and the physical position; Initializing a mechanical state memory in each virtual quantization unit, wherein the mechanical state memory comprises three storage areas, the first storage area is used for storing the axial stress component value of the virtual quantization unit at the current moment, the second storage area is used for storing the radial stress component value of the virtual quantization unit at the current moment, and the third storage area is used for storing the shear strain component value of the virtual quantization unit at the current moment; initializing a response sensitivity coefficient matrix inside each virtual quantization unit, wherein the response sensitivity coefficient matrix comprises a temperature disturbance response coefficient and a strain disturbance response coefficient, the temperature disturbance response coefficient represents the equivalent stress variation caused by unit temperature change at the virtual quantization unit, and the strain disturbance response coefficient represents the equivalent stress variation caused by unit mechanical strain at the virtual quantization unit; A state evolution rule function library is built in each virtual quantification unit, the state evolution rule function library comprises a stress relaxation evolution function and a stress accumulation evolution function, the stress relaxation evolution function calculates stress attenuation amount at the next moment according to the current stress state and the material viscoelasticity parameter, and the stress accumulation evolution function calculates strain amount increment at the next moment according to the current strain state and the external excitation intensity; based on the virtual quantization unit for completing the core attribute configuration, configuring a boundary interaction interface and a time sequence control module, assembling a chained model and completing parameter initialization and consistency verification.
3. 3. The intelligent prediction method for optical cable performance of data center according to claim 2, wherein the configuration boundary interaction interface and timing control module assemble a chain model and complete parameter initialization and consistency verification, comprising: Setting a boundary condition interface for each virtual quantization unit, wherein the boundary condition interface is used for receiving stress strain state information transmitted by adjacent virtual quantization units, the boundary condition interface comprises an input end and an output end, the input end receives stress state information of the adjacent virtual quantization units on the left side, and the output end transmits the stress state information of the unit to the adjacent virtual quantization units on the right side; A time step control module is configured in each virtual quantization unit, and the time step control module controls the calling frequency of a state evolution rule function library according to preset simulation time step parameters, wherein the simulation time step parameters are consistent with the acquisition time interval of the Brillouin scattering spectrum frequency shift data stream; Initializing and assigning an initial stress strain state parameter memory in each virtual quantification unit, wherein the initial values of the axial stress components of all the virtual quantification units are set to be prefabricated stress values when the optical cable leaves the factory, the initial values of the radial stress components are set to be equivalent stress values corresponding to the ambient atmospheric pressure, and the initial values of the shearing strain components are set to be zero; And connecting all the virtual quantization units in series according to the sequence of the spatial index identifiers, connecting boundary condition interfaces between adjacent virtual quantization units to form a chained virtual quantization unit sequence model, performing consistency check on the built chained virtual quantization unit sequence model, checking whether a response sensitivity coefficient matrix of each virtual quantization unit is in a preset reasonable range, checking whether the connection of the boundary condition interfaces of the adjacent virtual quantization units is correct, completing the check, solidifying model parameters and preparing to receive external input data.
4. 4. The intelligent prediction method for optical cable performance of a data center according to claim 1, wherein the inputting the optical cable acquired in real time into the optical cable microstructure disturbance quantization model along the brillouin scattering spectrum frequency shift data stream to perform disturbance source analysis processing, generating a physical excitation decomposition sequence to which each virtual quantization unit is subjected at the current moment, includes: Receiving an original data stream of a brillouin scattering spectrum frequency shift along an optical cable continuously acquired by a brillouin optical time domain reflectometer, wherein the original data stream comprises brillouin spectrum curves at the position of each spatial sampling point, and each brillouin spectrum curve consists of a frequency value and a corresponding scattered light intensity value; Carrying out peak detection processing on the Brillouin spectrum curve at each spatial sampling point position, and finding out a frequency value corresponding to the maximum value of scattered light intensity as the Brillouin frequency shift center frequency at the spatial sampling point position to obtain a Brillouin frequency shift center frequency curve longitudinally distributed along the optical cable; Reading a reference Brillouin frequency shift center frequency curve of the optical cable in a state without external disturbance from a pre-stored optical cable reference database, and subtracting the reference Brillouin frequency shift center frequency curve point by point from the Brillouin frequency shift center frequency curve acquired in real time to obtain a Brillouin frequency shift variation curve longitudinally distributed along the optical cable; The Brillouin frequency shift change amount curve is mapped in a segmented mode according to the spatial index identifiers of the virtual quantization units, each virtual quantization unit corresponds to a section of optical cable physical interval, and the average value of the Brillouin frequency shift change amounts of all spatial sampling points in the optical cable physical interval is taken as the integral Brillouin frequency shift change amount representative value of the virtual quantization unit; According to the linear coupling relation between the Brillouin frequency shift variable quantity and the temperature and the strain, establishing a frequency shift variable quantity decomposition equation of each virtual quantization unit, wherein the frequency shift variable quantity decomposition equation comprises a first frequency shift component caused by temperature disturbance and a second frequency shift component caused by strain disturbance, and the sum of the two components is equal to the integral Brillouin frequency shift variable quantity representative value of the virtual quantization unit; Invoking a response sensitivity coefficient matrix of each virtual quantization unit internal package, extracting a temperature disturbance response coefficient and a strain disturbance response coefficient of the virtual quantization unit from the response sensitivity coefficient matrix, wherein the temperature disturbance response coefficient represents the brillouin frequency shift change quantity caused by unit temperature change, and the strain disturbance response coefficient represents the brillouin frequency shift change quantity caused by unit strain change; Combining the frequency shift variation decomposition equation of each virtual quantization unit with a response sensitivity coefficient matrix to construct a binary primary equation set containing two unknowns, wherein the two unknowns are the temperature variation and the strain variation received by the virtual quantization unit at the current moment; solving a binary once equation set of each virtual quantization unit to obtain a temperature variation value and a strain variation value of the virtual quantization unit at the current moment, and forming a physical excitation decomposition result of the virtual quantization unit; and arranging physical excitation decomposition results of all the virtual quantization units at the current moment into a physical excitation decomposition sequence at the current moment according to the sequence of the space index identifiers from small to large, wherein each element in the physical excitation decomposition sequence corresponds to one virtual quantization unit and comprises the temperature change amount and the strain change amount of the virtual quantization unit.
5. 5. The intelligent prediction method for optical cable performance of data center according to claim 1, wherein the driving the virtual quantization units in the optical cable microstructure disturbance quantization model according to the physical excitation decomposition sequence to perform state evolution iterative operation, to obtain a stress-strain state evolution track set of each virtual quantization unit on a continuous time axis comprises: Reading values in a stress-strain state parameter memory at the current moment of all virtual quantification units from the optical cable microstructure disturbance quantization model, wherein the values comprise an axial stress component current value, a radial stress component current value and a shear strain component current value of each virtual quantification unit, and forming a stress-strain state matrix at the current moment; Inputting a physical excitation decomposition sequence at the current moment into a state evolution rule function library of each virtual quantization unit as an external driving force, and for each virtual quantization unit, respectively calculating the temperature change and the strain change in the physical excitation decomposition result with the corresponding coefficient in a response sensitivity coefficient matrix of the virtual quantization unit to obtain a stress increment predicted value of the virtual quantization unit in the next time step, wherein the stress increment predicted value is generated by external excitation; invoking a stress relaxation evolution function in a state evolution rule function library of each virtual quantization unit, inputting an axial stress component current value and a radial stress component current value at the current moment of the virtual quantization unit and a material viscoelasticity parameter, and calculating a stress attenuation predicted value at the next moment caused by a relaxation effect inside the material; Invoking a strain accumulation evolution function in a state evolution rule function library of each virtual quantization unit, inputting a current value of a shear strain component at the current moment of the virtual quantization unit and stress state information received from a boundary condition interface of an adjacent virtual quantization unit, and calculating a strain amount increment estimated value at the next moment caused by mechanical interaction of the adjacent units; Subtracting the stress attenuation amount predicted value from the stress increment predicted value of each virtual quantization unit to obtain an axial stress component updated value and a radial stress component updated value at the next moment of the virtual quantization unit, and adding the strain amount increment predicted value to the current value of the shear strain component to obtain a shear strain component updated value at the next moment of the virtual quantization unit; based on the single-time-step stress-strain updating predicted value, the stress interaction and boundary correction between the units are executed, the multi-time-step iterative operation is completed, and a stress-strain state evolution track set is generated.
6. 6. The intelligent prediction method for optical cable performance of data center according to claim 5, wherein the performing the inter-unit stress interaction and boundary correction, performing the multi-time-step iterative operation and generating the stress-strain state evolution track set comprises: Transmitting the updated axial stress component updated value of the unit to the adjacent virtual quantization unit on the right side through the boundary condition interface of each virtual quantization unit, and simultaneously receiving the updated axial stress component updated value of the adjacent virtual quantization unit on the left side from the boundary condition interface of the adjacent virtual quantization unit, so as to realize the interactive transmission of the stress states among the units; According to the received axial stress component updated value transmitted by the left adjacent virtual quantization unit, recalculating the boundary stress gradient of each virtual quantization unit, wherein the boundary stress gradient is the difference value between the axial stress component updated value of the left adjacent unit and the axial stress component updated value of the unit divided by the length of the basic structural unit; Feeding back the boundary stress gradient obtained by recalculation to the strain accumulation evolution function of each virtual quantification unit, and correcting the strain increment estimated value of the next time step to obtain a corrected shear strain component updated value; Writing the updated axial stress component updating values, the updated radial stress component updating values and the updated shear strain component updating values of all the virtual quantization units back to the corresponding stress strain state parameter memories, covering the original current moment value, and completing state evolution iteration of a time step; And sequentially driving physical excitation decomposition sequences corresponding to a plurality of continuous sampling time points, storing stress strain state parameters of all virtual quantization units at corresponding moments into a history record after each iteration of a time step is completed, and finally forming a stress tensor time-varying curve set and a strain tensor time-varying curve set of each virtual quantization unit on a continuous time axis.
7. 7. The intelligent prediction method for optical cable performance of a data center according to claim 1, wherein the constructing a space-time correlation network for optical cable link disturbance propagation based on the stress-strain state difference between adjacent virtual quantization units in the stress-strain state evolution track set comprises: extracting axial stress component time-varying curves of each virtual quantization unit on a complete time axis from the stress strain state evolution track set to obtain a group of axial stress time sequence curve sequences with ordered spatial positions, wherein the sequence of the curves in the sequence is completely consistent with the sequence of the spatial index identifiers of the virtual quantization units; For each pair of adjacent virtual quantization units, subtracting the axial stress time sequence curve of the left unit from the axial stress time sequence curve of the right unit from time points to obtain a time sequence change curve of the axial stress difference between the two units, wherein the time sequence change curve reflects the transmission effect of the stress state in the process of transmitting the disturbance from the left unit to the right unit; calculating the first derivative of each adjacent unit on the time dimension of the time sequence change curve of the axial stress difference to obtain a time sequence curve of the change rate of the axial stress difference, wherein the change rate of the axial stress difference represents the evolution speed of the stress state difference between the two units along with time, namely the instantaneous speed of disturbance transmission; taking each time point value on the corresponding axial stress difference change rate time sequence curve of each adjacent unit pair as an instantaneous weight value of the node connecting edge at the corresponding moment, taking a virtual quantization unit as a network node, taking directed connection between adjacent units as a network edge, marking the attribute of the edge by the instantaneous weight value, and constructing a time-space correlation network snapshot sequence which dynamically evolves along with time; For the time-space association network snapshot of each time point, the space index identifier of the virtual quantization unit is used as the coordinates of network nodes, the network nodes are arranged according to one-dimensional straight lines, and the connecting edges between the nodes only exist between the nodes of adjacent indexes, so that a weighting network with a chained topological structure is formed; Based on the time-space correlation network snapshot sequence, performing network weight normalization and smooth optimization to generate a final time-space correlation network for optical cable link disturbance propagation.
8. 8. The intelligent prediction method for optical cable performance of a data center according to claim 1, wherein the performing network flow characteristic analysis processing on the space-time correlation network to extract a set of early warning nodes in which stress strain states in network nodes accumulate beyond a critical threshold value comprises: Extracting axial stress component time-varying curves of each virtual quantization unit node at all time points from the time-space correlation network to obtain a stress history track of each node, wherein the stress history track records all stress state changes of the node from the initial moment to the current moment; performing accumulated damage calculation on stress history tracks of each node, identifying and extracting stress circulation in the tracks by adopting a rain flow counting method, and counting the times of stress circulation of different magnitudes, which are experienced by each node in a history time window, to form a stress circulation amplitude distribution histogram; According to the fatigue characteristic curve of the optical cable material, mapping each stress cycle amplitude in the stress cycle amplitude distribution histogram of each node into microscopic damage amount caused by the cycle to the material, and accumulating the microscopic damage amounts of all the stress cycles to obtain an accumulated fatigue damage value of the node; presetting a critical damage threshold value for each virtual quantification unit node, wherein the critical damage threshold value is determined by the ultimate tensile strength of the optical cable material and a design safety coefficient, and comparing the accumulated fatigue damage value of each node with the critical damage threshold value of the node; When the accumulated fatigue damage value of a certain virtual quantification unit node exceeds the critical damage threshold value, marking the node as a potential risk node, and recording a spatial index identifier of the node, the current accumulated fatigue damage value and a specific time point exceeding the threshold value; Performing space aggregation analysis on all virtual quantification units marked as potential risk nodes, detecting whether a plurality of potential risk nodes continuously adjacent in space position form a risk node cluster, and marking the whole cluster as an accumulated damage exceeding section if the number of the continuous adjacent nodes exceeds a preset cluster scale threshold; Extracting the time-varying curves of the shear strain components of all nodes in each accumulated damage exceeding section from the time-space correlation network, and calculating the slope of the change rate of each curve in the latest time window, wherein the slope of the change rate reflects the deterioration speed of the strain state of the node, and the larger the value of the slope of the change rate indicates the more severe the strain state is deteriorated; sequencing the change rate slopes of all nodes in the accumulated damage exceeding section, and selecting the first plurality of nodes with the largest change rate slope values as key early warning nodes, wherein the key early warning nodes represent local positions with the most severe strain state deterioration in the accumulated damage exceeding section; Mapping the spatial index identifier of each key early-warning node back to the actual physical position coordinate of the optical cable to obtain the actual geographic position information of the key early-warning node, wherein the actual geographic position information comprises the distance between the key early-warning node and the starting end of the optical cable and the specific laying environment section; And combining the spatial index identifier, the actual geographic position information, the current accumulated fatigue damage value and the strain state change rate slope of the key early-warning nodes to form an early-warning node information item, wherein all the early-warning node information items jointly form an early-warning node set.
9. 9. An intelligent prediction system for optical cable performance of a data center, comprising: A processor; a machine-readable storage medium storing machine-executable instructions for the processor; Wherein the processor is configured to perform the intelligent prediction method of cable performance for a data center of any one of claims 1 to 8 via execution of the machine executable instructions.
10. 10. A computer program product, characterized in that the computer program product comprises machine executable instructions stored in a computer readable storage medium, from which a processor of a smart prediction system for cable performance of a data center reads, the processor executing the machine executable instructions such that the smart prediction system for cable performance of a data center performs the smart prediction method for cable performance of a data center as claimed in any one of claims 1 to 8.

Description

Intelligent prediction method and system for optical cable performance of data center Technical Field The invention relates to the technical field of data center performance analysis, in particular to an intelligent optical cable performance prediction method and system for a data center. Background The data center has a complex environment, and the optical cable is affected by various external factors, such as temperature change, mechanical stress and the like, which may cause micro disturbance to the internal structure of the optical cable, further affect the transmission performance of the optical cable, and even cause data transmission faults. Currently, monitoring of the performance of optical cables is mainly dependent on traditional detection methods, such as periodic manual inspection and simple on-line monitoring equipment. The simple on-line monitoring equipment can only obtain certain single parameters of the optical cable, such as optical power and the like, and can not comprehensively and deeply know the stress-strain state of the internal structure of the optical cable and the propagation condition of disturbance. In addition, the existing monitoring method lacks the capability of predicting the abnormal performance of the optical cable, and can send out an alarm when a fault is generated or is about to happen, so that measures cannot be taken in advance for prevention, and potential risks are brought to the data transmission safety of the data center. Disclosure of Invention In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides an intelligent prediction method for optical cable performance of a data center, where the method includes: Constructing an optical cable microstructure disturbance quantization model, wherein the optical cable microstructure disturbance quantization model comprises a plurality of virtual quantization units distributed along the longitudinal direction of an optical cable, each virtual quantization unit corresponds to a basic structural unit in the optical cable, and each virtual quantization unit is internally packaged with a real-time stress strain state parameter of the basic structural unit and a response sensitivity coefficient of the basic structural unit to external disturbance; Inputting the optical cable acquired in real time into the optical cable microstructure disturbance quantization model along the line Brillouin scattering spectrum frequency shift data stream to perform disturbance source analysis processing, and generating a physical excitation decomposition sequence received by each virtual quantization unit at the current moment, wherein each element in the physical excitation decomposition sequence comprises a temperature disturbance component and a strain disturbance component which are obtained by analyzing the positions of the corresponding virtual quantization units; Driving virtual quantization units in the optical cable microstructure disturbance quantization model to perform state evolution iterative operation according to the physical excitation decomposition sequence to obtain a stress strain state evolution track set of each virtual quantization unit on a continuous time axis, wherein the stress strain state evolution track set comprises a stress tensor time-varying curve and a strain tensor time-varying curve of each virtual quantization unit; constructing a space-time correlation network for disturbance propagation of an optical cable link based on stress-strain state differences between adjacent virtual quantification units in the stress-strain state evolution track set, wherein the space-time correlation network takes the virtual quantification units as nodes, the change rate of the stress-strain state differences between adjacent nodes is taken as a weight value of node connecting edges, and the space-time correlation network is used for describing a transmission path and a transmission rate of the disturbance along the longitudinal direction of the optical cable; And carrying out network flow characteristic analysis processing on the space-time correlation network, extracting an early warning node set in which the stress strain state in the network nodes is accumulated to exceed a critical threshold, calculating the evolution trend of the stress strain state of each node in the early warning node set in a future time window, and generating serialized optical cable performance abnormality early warning information containing early warning node position coordinates and a stress strain state predicted value according to the evolution trend. In still another aspect, an embodiment of the present invention further provides an intelligent optical cable performance prediction system for a data center, including: the system comprises a processor, a machine-readable storage medium for storing machine-executable instructions of the processor, wherein