A charging pile life prediction system and its method

Abstract

The present invention discloses a charging pile life prediction system and method, which relates to the technical field of new energy equipment monitoring. It includes collecting multi-dimensional data signals during the operation of the charging pile, performing time alignment according to a preset data synchronization time window, and constructing a complete life characteristic index system; using wavelet transform to analyze the degradation trends of various signals, combining deep learning algorithms to perform spatio-temporal alignment and feature fusion on the collected multi-source heterogeneous data, establishing a degradation rate model to achieve feature extraction and health assessment of the life state; performing life feature extraction based on the results output by the degradation rate model, constructing a life state assessment model to achieve dynamic prediction and regular correction of the remaining life of the charging pile, and outputting optimization suggestions for equipment use according to the analysis results of life influencing factors. The present invention not only overcomes the prediction deviation problem caused by traditional single data sources, but also realizes all-round and multi-dimensional monitoring and evaluation of the health state of the charging pile.

Description

Technical Field

[0001] The present invention relates to the technical field of new energy equipment monitoring, and particularly to a charging pile life prediction system and method thereof. Background Art

[0002] With the booming development of the new energy vehicle industry, the construction scale of charging infrastructure has been continuously expanding, the scale of the charging network has been rapidly expanding, and the service life of equipment has been increasing, which puts forward higher requirements for the reliability and service life of charging piles. The health status monitoring and life prediction of charging facilities have become key technical issues to ensure the stable operation of the charging network. However, the existing health status monitoring of charging piles mainly relies on a single data source for evaluation, such as only collecting electrical parameters or temperature parameters. This single-dimensional monitoring method cannot comprehensively reflect the overall health status of the equipment, easily forms monitoring blind spots, and ultimately leads to insufficient accuracy and reliability of the evaluation results.

[0003] Currently, most of the evaluation methods adopt discriminant methods based on fixed thresholds and cannot adapt to the dynamic characteristics of the equipment under different working conditions. Especially in extreme working conditions such as high-power fast charging, the deterioration rate of the equipment may change suddenly, while the fixed evaluation method reacts slowly to this sudden change state and lacks necessary dynamic adjustment and optimization mechanisms. At the same time, due to the lack of effective processing methods for multi-source heterogeneous data, it is difficult for the existing technology to achieve the deep integration of multi-dimensional data such as acoustics, thermal imaging, and electromagnetic fields, and cannot make full use of the state information contained in various sensing data, which affects the accuracy of the prediction results. In addition, the current prediction models rarely consider the dynamic characteristics of equipment deterioration, and once the models are established, they are rarely updated, and cannot adapt to the dynamic changes of equipment status and usage environment, resulting in deviations between the prediction results and the actual situation. The existing technology also mainly focuses on fault diagnosis and emergency handling, lacks a predictive maintenance plan for the whole life cycle, and is difficult to achieve economic and efficient operation and maintenance of charging facilities. Summary of the Invention

[0004] In view of the problems existing in the existing charging pile life prediction methods, the present invention proposes a charging pile life prediction system and method thereof.

[0005] Therefore, the problem to be solved by the present invention lies in the lack of a predictive maintenance plan for the whole life cycle, which is difficult to achieve economic and efficient operation and maintenance of charging facilities.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a method for predicting the service life of a charging pile, which includes collecting multi-dimensional data signals during the operation of the charging pile, performing time alignment according to a preset data synchronization time window, and constructing a complete service life characteristic index system; analyzing the degradation trends of various signals using wavelet transform, and combining deep learning algorithms to perform spatio-temporal alignment and feature fusion on the collected multi-source heterogeneous data, establishing a degradation rate model, and realizing feature extraction and health assessment of the service life state; performing service life feature extraction based on the results output by the degradation rate model, constructing a service life state assessment model, and outputting optimization suggestions for device use according to the analysis results of service life influencing factors.

[0008] As a preferred solution of the method for predicting the service life of the charging pile according to the present invention, wherein: the multi-dimensional data signals include vibration spectrum signals, temperature distribution signals, electromagnetic field intensity signals, and electrical parameter signals, and the electrical parameter signals include power fluctuation signals and voltage and current signals.

[0009] As a preferred solution of the method for predicting the service life of the charging pile according to the present invention, wherein: performing discrete wavelet transform on the vibration spectrum signal to obtain a vibration characteristic frequency band, and extracting vibration amplitude data based on the vibration characteristic frequency band, which is expressed as:

[0010] ,

[0011] ϕ j , k [ n ] = 2 − j 2 ϕ ( 2 − j n − k ) ,

[0012] wherein, is the decomposition scale, is the translation parameter, x [ n ] is a discrete time series, ϕ j , k [ n ] is a discrete wavelet basis function, which is obtained by performing scale transformation and translation on the mother wavelet function; is the mother wavelet function, is the index of the discrete time series; the calculation of the vibration amplitude is expressed as:

[0013] ,

[0014] wherein, is the number of sampling points, is the vibration amplitude data, is the sampling point serial number; based on the temperature distribution signal, a temperature gradient matrix is established, and temperature stress data is extracted from the temperature gradient matrix. The calculation of the temperature gradient is expressed as:

[0015] ∇ T = [ ∂ T ∂ x , ∂ T ∂ y ] ,

[0016] wherein, represents the temperature gradient vector, represents the directional temperature change rate, represents the directional temperature change rate; The calculation of the temperature stress is expressed as:

[0017] ,

[0018] wherein, is the thermal stress value, is the material elastic modulus, is the linear thermal expansion coefficient, is the temperature change; Based on the electromagnetic field strength signal, electromagnetic field distortion data is obtained, and an electromagnetic feature vector is formed in combination with the electromagnetic field distortion data. The calculation of the electromagnetic field distortion is expressed as:

[0019] ,

[0020] wherein, is the electromagnetic field distortion rate, the th harmonic magnetic field strength, is the fundamental wave magnetic field strength, is the highest harmonic order considered; The power fluctuation signal and the voltage - current signal are converted into an electrical feature sequence, and power factor data and insulation resistance data are extracted from the electrical feature sequence. The calculation of the power factor is expressed as:

[0021] ,

[0022] wherein, is the power factor, is the active power, is the reactive power, is the apparent power; The determination of the insulation resistance is expressed as:

[0023] ,

[0024] wherein, is the insulation resistance value, is the test voltage, is the leakage current.

[0025] As a preferred scheme of the charging pile life prediction method described in the present invention, wherein: The time window is set as:

[0026] ,

[0027] wherein, is the time window length, is the sampling frequency of the vibration signal, is the sampling frequency of the temperature signal, is the sampling frequency of the electromagnetic field, is the sampling frequency of the electrical parameters, is the window expansion coefficient; The time alignment adopts the linear interpolation algorithm, expressed as:

[0028] ,

[0029] wherein, is the data value after interpolation, represents the data values of adjacent sampling points, is the interpolation time point, is the adjacent sampling time point.

[0030] As a preferred scheme of the charging pile life prediction method described in the present invention, wherein: based on the wavelet transform, time-frequency analysis is performed on the multi-source signals of vibration, temperature, and electromagnetic field, and the degradation characteristics of various signals are extracted. The basic formula of the continuous wavelet transform is expressed as:

[0031] ,

[0032] wherein, is the continuous wavelet transform coefficient matrix, is the continuous time signal, is the conjugate complex number of the continuous wavelet basis function, is the continuous wavelet basis function, is the continuous scale parameter, controlling the stretching degree of the wavelet; is the time parameter, controlling the translation position of the wavelet, representing the time position; is the energy normalization factor; Based on the time-frequency analysis, the time-frequency energy density is calculated, expressed as:

[0033] ,

[0034] wherein, is the time-frequency energy density distribution; Extract the degradation characteristics and analyze the degradation trend, expressed as:

[0035] ,

[0036] wherein, is the degradation characteristic sequence of the th type of signal, is the number of selected frequency scales, is the time point, is the th scale weight coefficient, is the Energy values at different scales; spatio-temporal alignment of multi-source heterogeneous data is achieved through an attention mechanism and a deep neural network. The attention mechanism includes a temporal attention mechanism and a spatial attention mechanism. The temporal attention mechanism is expressed as:

[0037] ,

[0038] where, is the attention weight in the time dimension, represents the time query transformation function, represents the time key transformation function, is the feature at the -th time point, is the feature at the -th time point, is the feature at the -th time point, is the length of the time series; the spatial attention mechanism is expressed as:

[0039] ,

[0040] where, is the attention weight in the spatial dimension, represents the spatial query transformation function, represents the spatial key transformation function, is the data of the -th feature dimension, is the data of the -th feature dimension, is the data of the -th feature dimension, is the number of feature dimensions.

[0041] As a preferred scheme of the charging pile life prediction method of the present invention: spatio-temporal alignment and feature fusion are performed based on the attention mechanism. The loss function of the spatio-temporal alignment is expressed as:

[0042] ,

[0043] ,

[0044] ,

[0045] where, is the total alignment loss, is the time alignment loss, is the spatial alignment loss, is the balance factor, is the feature of the -th signal at time , is the distribution of the th signal in the feature dimension , is the reference feature distribution, is the number of signals, is the number of feature dimensions; The formula for the feature fusion is expressed as:

[0046] ,

[0047] ,

[0048] wherein, is the initial fusion feature, is the final fusion feature, is the fusion weight, is the hidden layer representation of various features, is the feature fusion operator, is the attention enhancement function.

[0049] As a preferred solution of the charging pile life prediction method described in the present invention, wherein: The deterioration rate model is expressed as:

[0050] ,

[0051] wherein, is the deterioration rate of the charging pile at time , is the feature sensitivity parameter for controlling the feature influence degree, is the feature vector after multi-modal fusion, is the th state quantity of the monitoring signal, is the number of state variables, is the cumulative distribution function of the standard normal distribution for normalizing the output, is the comprehensive measured value of the environmental parameters, is the system state standard deviation; Based on the result output by the deterioration rate model, life feature extraction is performed to construct a life state evaluation model, and the formula is expressed as:

[0052] ,

[0053] wherein, is the health state index at time, is the deterioration rate, is the change in the deterioration rate, that is, , is the deterioration rate threshold, is the sigmoid function for normalization processing.

[0054] In a second aspect, an embodiment of the present invention provides a charging pile life prediction system, which includes: a collection module, configured to collect multi-dimensional data signals during the operation of the charging pile, perform time alignment according to a preset data synchronization time window, and construct a complete life characteristic index system; a prediction module, which uses wavelet transform to analyze the degradation trend of various signals, combines deep learning algorithms to perform spatio-temporal alignment and feature fusion on the collected multi-source heterogeneous data, establishes a degradation rate model, and realizes feature extraction and health assessment of the life state; an optimization module, which performs life feature extraction based on the results output by the degradation rate model, constructs a life state assessment model, and outputs optimization suggestions for device use according to the analysis results of life influencing factors.

[0055] In a third aspect, an embodiment of the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the processor executes the computer program, any step of the above-mentioned charging pile life prediction method is implemented.

[0056] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by a processor, any step of the above-mentioned charging pile life prediction method is implemented.

[0057] The beneficial effects of the present invention are as follows: By performing spatio-temporal alignment and feature fusion processing on the collected multi-source heterogeneous data, the present invention establishes a parameter degradation rate model, realizes feature extraction and health assessment of the life state, improves the accuracy and reliability of life prediction, overcomes the prediction deviation problem caused by a single data source, and realizes all-round monitoring and evaluation of the health state of the charging pile. It not only overcomes the prediction deviation problem caused by traditional single data sources, but also realizes all-round and multi-dimensional monitoring and evaluation of the health state of the charging pile. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained according to these drawings, where:

[0059] Figure 1 It is a scenario diagram of the charging pile life prediction method. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0060] To make the above objects, features, and advantages of the present invention more apparent and understandable, the following provides a detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0061] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0062] Secondly, the so-called "one embodiment" or "embodiment" herein refers to specific features, structures, or characteristics that may be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or selectively exclusive embodiments from other embodiments.

[0063] Example 1. Refer to Figure 1 , which is the first embodiment of the present invention. This embodiment provides a method for predicting the lifespan of a charging pile, including:

[0064] S1: Collect multi-dimensional data during the operation of the charging pile, including vibration spectrum, temperature distribution, electromagnetic field intensity, and power fluctuation parameters, and construct a complete lifespan characteristic index system.

[0065] By setting an acoustic sensor, an infrared thermal imager, an electromagnetic field sensor, and an electrical parameter acquisition device, collect the vibration spectrum signal, temperature distribution signal, electromagnetic field intensity signal, and electrical parameter signal of the charging pile. Among them, the electrical parameter signal includes a power fluctuation signal and a voltage and current signal.

[0066] Set the sampling frequency of the vibration spectrum signal to 1 kHz, the sampling frequency of the temperature distribution signal to 10 Hz, the sampling frequency of the electromagnetic field intensity signal to 500 Hz, and the sampling frequency of the electrical parameter signal to 50 Hz.

[0067] Perform discrete wavelet transform on the vibration spectrum signal to obtain the vibration characteristic frequency band, and extract the vibration amplitude data based on the vibration characteristic frequency band. The discrete wavelet transform formula is expressed as:

[0068] DWT ( j , k ) = ∑ n x [ n ] ⋅ ϕ j , k [ n ] ,

[0069] ϕ j , k [ n ] = 2 − j 2 ϕ ( 2 − j n − k ) ,

[0070] Wherein, is the decomposition scale, is the translation parameter, x [ n ] is a discrete time series, ϕ j , k [ n ] is a discrete wavelet basis function, obtained by scaling and translating the mother wavelet function; is the mother wavelet function, is the index of the discrete time series.

[0071] The calculation of the vibration amplitude is expressed as:

[0072] ,

[0073] where, is the number of sampling points, is the vibration amplitude data, is the sampling point number.

[0074] The temperature distribution signal is obtained through infrared thermal imaging, a temperature gradient matrix is established, and the temperature stress data of key components are extracted from the temperature gradient matrix. The calculation of the temperature gradient is expressed as:

[0075] ∇ T = [ ∂ T ∂ x , ∂ T ∂ y ] ,

[0076] where, represents the temperature gradient vector, represents the temperature change rate in the direction, represents the temperature change rate in the

[0077] The temperature stress is further calculated:

[0078] ,

[0079] where, is the thermal stress value, is the material elastic modulus, is the linear thermal expansion coefficient, is the temperature change amount.

[0080] Based on the Hall sensor array, the electromagnetic field intensity signal is collected, the electromagnetic field distortion data is obtained, and the electromagnetic feature vector is formed by combining the electromagnetic field distortion data. The calculation of the electromagnetic field distortion is expressed as:

[0081] ,

[0082] where, is the electromagnetic field distortion rate, the th harmonic magnetic field intensity, is the fundamental magnetic field strength, is the highest harmonic order considered.

[0083] Convert the power fluctuation signal and voltage and current signals into an electrical feature sequence, and extract power factor data and insulation resistance data from the electrical feature sequence. The calculation of the power factor is expressed as:

[0084] ,

[0085] where, is the power factor, is the active power, is the reactive power, is the apparent power.

[0086] The determination of the insulation resistance is expressed as:

[0087] ,

[0088] where, is the insulation resistance value, is the test voltage, is the leakage current.

[0089] According to the preset data synchronization time window, perform time alignment on the vibration amplitude data, temperature stress data, electromagnetic feature vector, power factor data, and insulation resistance data. The time window is set as:

[0090] ,

[0091] where, is the time window length, is the vibration signal sampling frequency, is the temperature signal sampling frequency, is the electromagnetic field sampling frequency, is the electrical parameter sampling frequency, is the window expansion coefficient.

[0092] The time alignment adopts a linear interpolation algorithm, expressed as:

[0093] ,

[0094] where, is the interpolated data value, represents the data value of adjacent sampling points, is the interpolation time point, is the adjacent sampling time point.

[0095] The time-aligned data is composed into a feature data matrix, and the outliers in the feature data matrix are corrected to obtain the preprocessed life characteristic index data.

[0096] S2: Analyze the degradation trends of various signals using wavelet transform, perform spatio-temporal alignment and feature fusion on the collected multi-source heterogeneous data in combination with deep learning algorithms, establish a parameter degradation rate model, and realize the feature extraction and health assessment of the life state.

[0097] Perform time-frequency analysis on multi-source signals such as vibration, temperature, and electromagnetic field using continuous wavelet transform, and extract the degradation characteristics of various signals. The basic formula of continuous wavelet transform is expressed as:

[0098] ,

[0099] where is the continuous wavelet transform coefficient matrix, is the continuous-time signal, is the conjugate complex number of the continuous wavelet basis function, is the continuous wavelet basis function, is the continuous scale parameter, which controls the stretching degree of the wavelet; is the time parameter, which controls the translation position of the wavelet and represents the time position; is the energy normalization factor.

[0100] Based on time-frequency analysis, calculate the time-frequency energy density, which is expressed as:

[0101] ,

[0102] where is the time-frequency energy density distribution.

[0103] Extract the degradation characteristics, which are expressed as:

[0104] ,

[0105] where is the degradation characteristic sequence of the th type of signal, is the number of selected frequency scales, is the time point, is the th scale weight coefficient, is the th scale energy value.

[0106] Align the multi-source heterogeneous data through the attention mechanism and the deep neural network. The attention mechanism includes the time attention mechanism and the spatial attention mechanism. The time attention mechanism is expressed as:

[0107] ,

[0108] Among them, is the attention weight in the time dimension, represents the time query transformation function, represents the time key transformation function, is the feature at the -th time point, is the feature at the -th time point, is the feature at the -th time point, is the length of the time series.

[0109] By selecting the features at three different time points, calculating the change trend, capturing the dynamic response characteristics of the system, reflecting the evolution process of the system state, and providing the temporal context of the system behavior, it not only ensures sufficient temporal information but also avoids excessive computational complexity.

[0110] The spatial attention mechanism is expressed as:

[0111] ,

[0112] Among them, is the attention weight in the spatial dimension, represents the spatial query transformation function, represents the spatial key transformation function, is the data of the -th feature dimension, is the data of the -th feature dimension, is the data of the -th feature dimension, is the number of feature dimensions.

[0113] Then, spatio-temporal alignment is performed according to the time features and spatial features, and the spatio-temporal alignment loss function is expressed as:

[0114] ,

[0115] ,

[0116] ,

[0117] Among them, is the total alignment loss, is the time alignment loss, is the spatial alignment loss, is the balance factor, is the feature of the -th signal at time .​ is the distribution of the th signal in the feature dimension , is the reference feature distribution, is the number of signals, is the number of feature dimensions.

[0118] Align the features of different time series through the spatio-temporal alignment loss function, reduce the differences caused by time delay, make the features consistent in the time dimension, and capture similar time series patterns; map the features of different devices to the same distribution space, eliminate the systematic errors between devices, standardize the feature distribution, establish a unified measurement standard, obtain a time-synchronized feature sequence and a standardized feature space, reduce the impact of data deviation, and improve the prediction accuracy.

[0119] Fuse the transformed features of each type in a weighted combination manner according to the data after spatio-temporal alignment, assign corresponding fusion weights to each type of feature, and further enhance the expression ability of the features through the residual connection structure to ensure that the original information will not be lost during the fusion process.

[0120] The formula for feature fusion is expressed as:

[0121] ,

[0122] ,

[0123] where, is the initial fusion feature, is the final fusion feature, is the fusion weight, is the hidden layer representation of each type of feature, is the feature fusion operator, is the attention enhancement function.

[0124] Construct a degradation rate model through the data features after fusion to describe the performance degradation process of the charging pile. The formula is expressed as:

[0125] ,

[0126] where, is the degradation rate of the charging pile at time , is the feature sensitivity parameter used to control the influence degree of the feature, is the feature vector after multi-modal fusion, is the th state quantity of the monitoring signal, is the number of state variables, is the cumulative distribution function of the standard normal distribution for normalizing the output, is the comprehensive measurement value of environmental parameters, is the standard deviation of the system state.

[0127] S3: Extract life characteristics based on the results output by the degradation rate model, construct a life state evaluation model, and output optimization suggestions for equipment use according to the analysis results of life influencing factors.

[0128] Extract life characteristics according to the results output by the degradation rate model to achieve an accurate assessment of the health state of the charging pile. The formula of the life state evaluation model is:

[0129] ,

[0130] where, is the health state index at time is the degradation rate, is the change in the degradation rate, i.e., , is the degradation rate threshold, is the sigmoid function for normalization processing.

[0131] Characterize the degradation process of the charging pile through multi-source data fusion and cumulative effect analysis, achieve a quantitative assessment of the health state of the charging pile, timely detect the trend of equipment performance decline, and provide decision support for predictive maintenance.

[0132] Set threshold intervals based on the output results of the life state evaluation model, including the first threshold , the second threshold , the third threshold , the fourth threshold , and >[[]] >[[]] >[[]] .

[0133] When ≥ , it indicates that the equipment is in a stable operation stage;

[0134] When ≤ < , it indicates that the equipment is in the initial degradation stage;

[0135] When ≤ < , it indicates that the equipment is in the accelerated degradation stage;

[0136] When ≤ < When it indicates that the equipment is in the severe deterioration stage;

[0137] When < it indicates that the equipment is in the failure risk stage.

[0138] If the charging pile is in the stable operation stage, continue to execute the regular monitoring and maintenance plan;

[0139] If the charging pile is in the failure risk stage, it is necessary to immediately shut down and conduct a comprehensive overhaul;

[0140] If the charging pile is in the initial deterioration stage or the accelerated deterioration stage, enter the secondary assessment based on the deterioration rate to judge the degree of deterioration;

[0141] Set the threshold of the deterioration rate as , , and < ;

[0142] If the charging pile is in the initial deterioration stage and the deterioration rate < , it is determined to be in a slightly deteriorated state, record and analyze the equipment usage conditions, establish a deterioration characteristic database, formulate a preventive maintenance plan for the equipment, and prepare for subsequent maintenance;

[0143] If the charging pile is in the initial deterioration stage and the deterioration rate < < , it is determined to be in a moderately deteriorated state, start the preventive maintenance plan, arrange professional personnel for on-site inspection, conduct a comprehensive performance test on the charging module and control system, check and clean the heat dissipation system, connectors and other easily deteriorated components, evaluate the equipment operation environment, and optimize the usage strategy;

[0144] If the charging pile is in the initial deterioration stage and the deterioration rate < , it is determined to be in a rapidly deteriorated state, immediately reduce the equipment load, limit the high-power charging mode, analyze the reason for the rapid deterioration, check whether there are systematic problems, establish an emergency response mechanism, and prepare an alternative equipment plan.

[0145] If the charging pile is in the accelerated deterioration stage and the deterioration rate < , it is determined to be in a conventional accelerated deterioration state, conduct a comprehensive inspection and performance evaluation on the charging module, establish a real-time monitoring and early warning mechanism for the equipment status, prepare necessary replacement parts and maintenance tools, and formulate a detailed maintenance plan and emergency plan.

[0146] If the charging pile is in the accelerated deterioration stage and the deterioration rate < < If it is determined to be in a rapid acceleration deterioration state, the operating power of the device will be immediately reduced to 50% of the rated power, a intensive monitoring mechanism will be started once an hour to conduct stress tests and life assessments on key components, complete spare core components will be prepared, whether it is necessary to replace the entire machine equipment in advance will be evaluated, the root cause of the accelerated deterioration will be analyzed and a prevention mechanism will be established.

[0147] If the charging pile is in the stage of accelerated deterioration and the deterioration rate < If it is determined to be in a dangerous acceleration deterioration state, the operation of the device needs to be stopped immediately, the main circuit power supply needs to be cut off, alternative equipment needs to be prepared to ensure that the service is not interrupted, an emergency replacement plan will be formulated, including equipment, personnel, and tool configuration; a comprehensive safety hazard investigation will be carried out, preventive inspections will be carried out on similar equipment to prevent chain reactions, an accident cause analysis report will be established, and the prevention mechanism will be improved.

[0148] In summary, through multi-dimensional sensing devices, the present invention comprehensively collects key operation parameters of the charging pile during operation, constructs a complete life characteristic index evaluation system, innovatively introduces a deep learning algorithm to perform spatio-temporal alignment and feature fusion processing on the collected multi-source heterogeneous data, establishes a parameter deterioration rate model, realizes feature extraction and health assessment of the life state, improves the accuracy and reliability of life prediction, overcomes the prediction deviation problem caused by a single data source, and realizes all-round monitoring and evaluation of the health state of the charging pile. It not only overcomes the prediction deviation problem caused by traditional single data sources, but also realizes all-round and multi-dimensional monitoring and evaluation of the health state of the charging pile. The present invention provides reliable technical support for the predictive maintenance of charging infrastructure, and has important theoretical guiding significance and practical application value for improving the intelligent operation and maintenance level of charging facilities, extending the service life of equipment, and reducing maintenance costs.

[0149] This embodiment further provides a charging pile life prediction system, including:

[0150] A collection module, which is used to collect multi-dimensional data signals during the operation of the charging pile, perform time alignment according to a preset data synchronization time window, and construct a complete life characteristic index system;

[0151] A prediction module, which uses wavelet transform to analyze the degradation trend of various signals, combines a deep learning algorithm to perform spatio-temporal alignment and feature fusion on the collected multi-source heterogeneous data, establishes a deterioration rate model, and realizes feature extraction and health assessment of the life state;

[0152] An optimization module extracts life characteristics based on the results output by the degradation rate model, constructs a life state evaluation model, realizes dynamic prediction and regular correction of the remaining life of the charging pile, and outputs optimization suggestions for equipment use according to the analysis results of life influencing factors.

[0153] This embodiment also provides a computer device applicable to the case of the charging pile life prediction method, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the charging pile life prediction method proposed in the above embodiment.

[0154] This computer device can be a terminal, and this computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of this computer device is used to provide computing and control capabilities. The memory of this computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of this computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a carrier network, NFC (Near Field Communication), or other technologies. The display screen of this computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of this computer device can be a touch layer covered on the display screen, or a button, a trackball, or a touchpad set on the shell of the computer device, or an external keyboard, touchpad, or mouse, etc.

[0155] This embodiment also provides a storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the charging pile life prediction method proposed in the above embodiment.

[0156] The storage medium proposed in this embodiment and the data storage method proposed in the above embodiment belong to the same inventive concept. The technical details not described in detail in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0157] Embodiment 2. This embodiment provides a charging pile life prediction method. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0158] To verify the effectiveness of the proposed method, 10 DC fast chargers with a running time of more than 3 years in a charging station are selected for an experimental study for 6 months. The rated power of these charging piles is 120kW, and the average daily usage duration is about 12 hours. The experiment compares the traditional single data evaluation method, the fixed threshold evaluation method with the method of the present invention.

[0159] Install acoustic sensors, infrared thermal imagers, electromagnetic field sensors, and electrical parameter acquisition devices on each charging pile. Set the sampling frequencies as follows: acoustic data at 100 Hz, thermal imaging at 30 fps, electromagnetic field data at 50 Hz, and electrical parameters at 1 Hz. Conduct a health status assessment and life prediction every 2 hours, using the results of manual inspection as the evaluation criteria.

[0160] The experimental results are shown in Table 1:

[0161] Table 1 Comparative Evaluation Table

[0162] Evaluation index Traditional single - data method Fixed - threshold method Method of the present invention Accuracy of state evaluation 76.3% 82.5% 94.8% Error of life prediction ±25.6% ±18.4% ±8.2% Advance amount of fault warning 24 hours 48 hours 96 hours False - alarm rate 15.3% 12.7% 4.2% Miss - alarm rate 12.8% 8.9% 2.1% Evaluation delay 5 minutes 8 minutes 12 minutes Recognition rate of abnormal working conditions 65.4% 77.2% 92.6%

[0163] The experimental data shows that in terms of the accuracy of status assessment, the method of the present invention reaches 94.8%, showing a significant improvement compared to 76.3% of the traditional single-data method and 82.5% of the fixed-threshold method; in terms of the life prediction error, the present invention controls the error within the range of ±8.2%, with an improvement of more than 50% compared to ±25.6% of the traditional method and ±18.4% of the fixed-threshold method.

[0164] In terms of the fault warning ability, the present invention can issue a warning 96 hours in advance, which is 4 times the 24-hour warning time of the traditional method, providing a more sufficient response time for equipment maintenance. At the same time, the method of the present invention reduces the false alarm rate and missed alarm rate to 4.2% and 2.1% respectively, achieving a qualitative leap compared to 15.3% and 12.8% of the existing technology.

[0165] Although the evaluation delay of the present invention is slightly higher than that of the existing method, reaching 12 minutes, considering its significantly improved prediction accuracy and reliability, this time cost is completely acceptable. Especially in the identification of abnormal working conditions, the identification rate of the method of the present invention is as high as 92.6%, far exceeding 65.4% of the traditional method and 77.2% of the fixed-threshold method.

[0166] In summary, the experimental results fully verify the technical advantages and application value of the present invention in the field of health status assessment and life prediction of charging piles.

[0167] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A charging pile life prediction method, characterized in that: include: Collect multi-dimensional data signals during the operation of the charging pile, perform time alignment according to the preset data synchronization time window, and build a complete life characteristic indicator system; Wavelet transform is used to analyze the degradation trends of various signals, and deep learning algorithms are used to align the time and space and fuse the features of the collected multi-source heterogeneous data, establish a degradation rate model, and realize feature extraction and health assessment of life status; Based on the output of the degradation rate model, life characteristics are extracted, and a life status assessment model is constructed. Based on the analysis results of life influencing factors, equipment usage optimization suggestions are output; The spatiotemporal alignment of multi-source heterogeneous data is achieved through the attention mechanism and deep neural network. The attention mechanism includes the temporal attention mechanism and the spatial attention mechanism. The temporal attention mechanism is expressed as: ， in, is the attention weight in the time dimension, represents the time query transformation function, represents the time key transformation function, For the The characteristics of a time point, For the The characteristics of a time point, For the The characteristics of a time point, is the length of the time series; The spatial attention mechanism is expressed as: ， in, is the attention weight of the spatial dimension, represents the spatial query transformation function, represents the spatial key transformation function, For the feature dimension data, For the feature dimension data, For the feature dimension data, is the number of feature dimensions; The degradation rate model is expressed as: ， in, For charging piles in time The degradation rate of is the feature sensitivity parameter used to control the degree of feature influence. is the feature vector after multimodal fusion, For the The state quantity of the monitoring signal, is the number of state variables, is the standard normal distribution cumulative function used to normalize the output, is the comprehensive measurement value of environmental parameters, is the system state standard deviation; Based on the output of the degradation rate model, life characteristics are extracted and a life status assessment model is constructed. The formula is expressed as follows: ， in, for The health status index at all times, is the degradation rate, is the change in degradation rate, , is the degradation rate threshold, The sigmoid function is used for normalization.

2. The charging pile life prediction method according to claim 1, characterized in that: The multi-dimensional data signal includes a vibration spectrum signal, a temperature distribution signal, an electromagnetic field strength signal and an electrical parameter signal, wherein the electrical parameter signal includes a power fluctuation signal and a voltage and current signal.

3. The charging pile life prediction method according to claim 2, characterized in that: The vibration spectrum signal is subjected to discrete wavelet transform to obtain a vibration characteristic frequency band, and vibration amplitude data is extracted based on the vibration characteristic frequency band, which is expressed as: ， ， in, To decompose the scale, is the translation parameter, is a discrete time series, is the discrete wavelet basis function, which is obtained by scaling and translating the mother wavelet function; is the mother wavelet function, is the index of the discrete time series; The calculation of the vibration amplitude is expressed as: ， in, is the number of sampling points, is the vibration amplitude data, is the sampling point number; Based on the temperature distribution signal, a temperature gradient matrix is ​​established, and temperature stress data is extracted from the temperature gradient matrix. The temperature gradient is calculated as follows: ， in, represents the temperature gradient vector, express Directional temperature change rate, express Directional temperature change rate; The calculation of the temperature stress is expressed as: ， in, is the thermal stress value, is the elastic modulus of the material, is the linear thermal expansion coefficient, is the temperature change; Based on the electromagnetic field strength signal, electromagnetic field distortion data is obtained, and the electromagnetic field distortion data is combined to form an electromagnetic feature vector, wherein the electromagnetic field distortion calculation is expressed as: ， in, is the electromagnetic field distortion rate, No. Subharmonic magnetic field strength, is the fundamental magnetic field strength, is the highest harmonic order considered; The power fluctuation signal and the voltage and current signal are converted into an electrical characteristic sequence, and power factor data and insulation resistance data are extracted from the electrical characteristic sequence, wherein the calculation of the power factor is expressed as: ， in, is the power factor, is the active power, is the reactive power, is the apparent power; The determination of the insulation resistance is expressed as: ， in, is the insulation resistance value, For the test voltage, is the leakage current.

4. The charging pile life prediction method according to claim 3, characterized in that: The time window is set to: ， in, is the time window length, is the vibration signal sampling frequency, is the temperature signal sampling frequency, is the electromagnetic field sampling frequency, is the electrical parameter sampling frequency, is the window expansion factor; The time alignment adopts a linear interpolation algorithm, which is expressed as: ， in, is the interpolated data value, Represents the data values ​​of adjacent sampling points, is the interpolation time point, are adjacent sampling time points.

5. The charging pile life prediction method according to claim 4, characterized in that: Based on the wavelet transform, the time-frequency analysis of multi-source signals of vibration, temperature and electromagnetic field is performed to extract the degradation characteristics of various signals. The basic formula of continuous wavelet transform is expressed as: ， in, is the continuous wavelet transform coefficient matrix, is a continuous time signal, is the conjugate complex number of the continuous wavelet basis function, is the continuous wavelet basis function, is a continuous scale parameter that controls the degree of expansion and contraction of the wavelet; is the time parameter, which controls the translation position of the wavelet and represents the time position; is the energy normalization factor; Based on time-frequency analysis, the time-frequency energy density is calculated and expressed as: ， in, is the time-frequency energy density distribution; Extract degradation features and analyze degradation trends, expressed as: ， in, For the The degenerate characteristic sequence of the signal-like is the number of frequency scales selected, For time point, For the The weight coefficient of each scale, For the The energy value of a scale.

6. The method for predicting the life of a charging pile according to claim 5, characterized in that: Based on the attention mechanism, spatiotemporal alignment and feature fusion are performed. The loss function of the spatiotemporal alignment is expressed as: ， ， ， in, is the total alignment loss, is the time alignment loss, is the spatial alignment loss, is the balance factor, For the Signal at time Features, For the The signal in the feature dimension The distribution of is the reference feature distribution, is the number of signals, is the number of feature dimensions; The formula of feature fusion is expressed as: ， ， in, is the initial fusion feature, is the final fusion feature, is the fusion weight, is the hidden layer representation of various features, is the feature fusion operator, is the attention enhancement function.

7. A charging pile life prediction system, based on the charging pile life prediction method according to any one of claims 1 to 6, characterized in that: include: The acquisition module is used to collect multi-dimensional data signals during the operation of the charging pile, perform time alignment according to the preset data synchronization time window, and build a complete life characteristic indicator system; The prediction module uses wavelet transform to analyze the degradation trends of various signals, combines deep learning algorithms to perform spatiotemporal alignment and feature fusion of the collected multi-source heterogeneous data, establishes a degradation rate model, and realizes feature extraction and health assessment of life status; The optimization module extracts life characteristics based on the output of the degradation rate model, builds a life status assessment model, and outputs equipment usage optimization suggestions based on the analysis results of life influencing factors.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the charging pile life prediction method described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the charging pile life prediction method described in any one of claims 1 to 6 are implemented.

Images