Method for monitoring direct arc current of optical storage power station

Abstract

The invention provides a direct arc current monitoring method for an optical storage power station. The method comprises the steps of 1, high-frequency sampling; 2, multi-feature extraction: respectively calculating a time domain feature kurtosis, a pulse factor and a current change rate (di / dt); 3-100 kHz frequency band energy distribution is extracted, and wavelet packet decomposition is adopted to obtain frequency domain characteristics of each sub-band energy ratio; 3, a dynamic threshold value model is established, wherein a threshold value correction formula is established based on the working voltage (vmp) of the photovoltaic array and the environment temperature (T); 4, improved D-S evidence theory fusion: introducing weighted credibility distribution, and dynamically updating the feature weight through Kalman filtering; relates to the technical field of safety monitoring of photovoltaic power stations and energy storage systems, solves the problems of high false alarm rate, long response time, incapability of distinguishing normal transient and real arcs, difficulty in adapting to complex working conditions and the like in the prior art, and realizes accurate and rapid detection of arc faults on the direct current side of an optical storage power station. And safe operation of the optical storage power station and the energy storage system is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of safety monitoring for photovoltaic power stations and energy storage systems, and particularly to a method for monitoring direct arc current in a photovoltaic and energy storage power station. Background Art

[0002] In photovoltaic power stations and energy storage systems, safety monitoring is crucial. Arc faults on the DC side of photovoltaic systems are the main cause of power station fires. However, traditional AC arc detection methods are not applicable to DC systems, and there are many deficiencies in the existing technologies:

[0003] Relying on a single frequency domain feature: Current detection methods mostly rely on a single frequency domain feature. For example, they only focus on the energy in the range of 3 - 5 kHz. However, in the actual operating environment, photovoltaic and energy storage power stations are extremely vulnerable to switching noise, electromagnetic interference, etc. This results in a false alarm rate of over 30%, seriously affecting the stability and reliability of the system, and increasing unnecessary maintenance costs and safety hazards.

[0004] Too long response time: The response time of existing technologies exceeds 100 ms, which cannot meet the 50 - ms fast - cut - off requirement of the IEC62548 standard. Once an arc fault occurs, the too - long response time may cause the fire to spread rapidly, resulting in irreparable losses.

[0005] Unable to effectively distinguish normal transients from real arcs: In actual operation, there are normal transient processes such as load switching and inverter start - stop in photovoltaic and energy storage power stations. Existing detection methods cannot accurately distinguish them from real arcs, resulting in frequent false alarms or missed alarms and affecting the normal operation of the system.

[0006] Complex operating conditions increase the detection difficulty: The charging and discharging operating conditions of photovoltaic and energy storage power stations are complex, such as maximum power point tracking (MPPT) and battery dynamic response. These complex operating conditions further increase the detection difficulty, and existing detection methods are difficult to adapt to such a complex environment.

[0007] Therefore, there is an urgent need for a method for monitoring direct arc current in a photovoltaic and energy storage power station that can effectively solve the above problems, in order to improve the safety and reliability of photovoltaic power stations and energy storage systems. Summary of the Invention

[0008] In order to overcome the existing problems, the embodiments of the present application provide a method for monitoring direct arc current in a photovoltaic and energy storage power station, so as to solve the problems of high false alarm rate, long response time, inability to distinguish normal transients from real arcs, and difficulty in adapting to complex operating conditions in the existing technologies, and achieve accurate and rapid detection of arc faults on the DC side of photovoltaic power stations, ensuring the safe operation of photovoltaic power stations and energy storage systems.

[0009] The technical solutions adopted by the embodiments of the present application to solve their technical problems are:

[0010] A method for monitoring direct arc current of a photovoltaic power station comprises the following steps:

[0011] Step 1: High-frequency sampling: Obtain the DC side current signal at a sampling rate of ≥100kHz, and use a high-precision current sensor to collect the current signal to ensure the accuracy and stability of the collected data;

[0012] Among them, the DC side current signal is obtained at a sampling rate of ≥100kHz. The high sampling rate can more accurately capture the subtle changes in the current signal, provide a richer and more accurate data basis for subsequent feature extraction, and effectively avoid information loss caused by insufficient sampling rate;

[0013] Step 2, multi-feature extraction: calculate the time domain feature kurtosis, pulse factor, and current change rate (di / dt) respectively; extract the energy distribution of the 3-100kHz frequency band, and use wavelet packet decomposition to obtain the frequency domain characteristics of the energy proportion of each sub-band; calculate the nonlinear characteristics of sample entropy, Hurst index, and Lyapunov index;

[0014] Among them, the time domain characteristics: calculate the kurtosis, pulse factor, and current change rate (di / dt). The kurtosis is used to measure the degree of deviation of the signal from the normal distribution and can highlight the impact component in the signal; the pulse factor is sensitive to the pulse characteristics in the signal and helps to detect the instantaneous pulses generated by the arc fault; the current change rate reflects how fast the current changes over time. When an arc fault occurs, the current change rate usually fluctuates abnormally;

[0015] Frequency domain features: Extract the energy distribution of the 3-100kHz frequency band, and use wavelet packet decomposition to obtain the energy proportion of each sub-band. This frequency band covers the main frequency components generated by arc faults. By analyzing the energy distribution and the energy proportion of each sub-band, we can more comprehensively understand the characteristics of the signal in different frequency bands and improve the accuracy of detection;

[0016] Nonlinear features: Calculate sample entropy, Hurst index, and Lyapunov index. These nonlinear features can mine complex nonlinear information in signals, effectively distinguish normal signals from arc fault signals, and enhance the detection capability of arc faults under complex working conditions.

[0017] Step 3: Dynamic threshold model: Based on the photovoltaic array operating voltage (v mp ), ambient temperature (T) to establish a threshold correction formula, which is derived based on a large amount of experimental data and theoretical analysis to achieve automatic optimization of detection sensitivity as the environment changes;

[0018] Among them, based on the photovoltaic array working voltage (v mp) Establish a threshold correction formula for the ambient temperature (T). The operating state of the photovoltaic and energy storage power station is greatly affected by environmental factors. Through this dynamic threshold model, the detection sensitivity can be automatically optimized according to environmental changes, ensuring accurate detection of arc faults under different environmental conditions;

[0019] Step 4. Improve the D-S evidence theory fusion: Introduce weighted credibility assignment. The weighted credibility assignment is initially assigned according to the importance of different features for arc fault detection under different working conditions, and the feature weights are dynamically updated through Kalman filtering; Use the Kalman filtering algorithm to dynamically update the feature weights to adapt to different operating conditions. When the fusion confidence level ≥ 95%, trigger protection, and the response time ≤ 30 ms;

[0020] Among them, introduce weighted credibility assignment and dynamically update the feature weights through Kalman filtering. Considering that different features have different contributions to arc fault detection under different working conditions, dynamically updating the feature weights can more reasonably fuse multi-feature information. Define the decision rule: When the fusion confidence level ≥ 95%, trigger protection, and the response time ≤ 30 ms, ensuring accurate judgment of arc faults while meeting the requirement of rapid cut-off and effectively reducing the accident risk.

[0021] Preferably, the straight arc current monitoring method further includes recording the time of fault occurrence, current signal characteristics, and environmental parameter information after triggering protection for subsequent analysis of the fault cause.

[0022] Preferably, during the entire monitoring process of the straight arc current monitoring method, the collected data, calculated feature values, and monitoring status are transmitted to the monitoring center through the communication module in real time to achieve remote monitoring and management.

[0023] Preferably, the time domain features include the kurtosis (K), pulse factor (PF), and current change rate (di / dt) of the current signal, where:

[0024]

[0025] μ is the current mean value, σ is the standard deviation, and I peak is the peak current.

[0026] Preferably, in the dynamic threshold model, the threshold correction formula based on the working voltage (v mp ) of the photovoltaic array and the ambient temperature (T):

[0027]

[0028] Among them, I0 is the reference threshold, and α, β are correction coefficients.

[0029] Preferably, the improved D-S evidence theory includes:

[0030] Initial credibility weights ω1, ω2, and ω3 are respectively assigned to the time domain, frequency domain, and non-linear features (Σwi = 1);

[0031] The weights are dynamically updated through a Kalman filter, and the update formula is:

[0032] ω i (k) = ω i (k - 1) + K k *[z k - h(ω i (k - 1))]

[0033] K k is the Kalman gain, z k is the k-th observation value, and h(ω i (k - 1)) is the observation function.

[0034] The advantages of the embodiments of the present application are:

[0035] In the present invention, by parallelly extracting time-domain, frequency-domain, and non-linear features and combining with the improved D-S evidence theory for multi-feature fusion decision-making, the limitations of single-feature detection are effectively avoided, the false alarm rate is greatly reduced, and the detection accuracy is improved; multi-dimensional feature extraction can comprehensively capture the features of current signals, enabling the system to accurately distinguish normal transient processes such as load switching and inverter start / stop from real arcs, avoiding false alarms and missed detections, and ensuring the normal operation of the system; the dynamic threshold model enables the detection sensitivity to be automatically optimized with environmental changes, and the multi-feature fusion method fully considers the complex characteristics of the charge and discharge conditions of the photovoltaic energy storage power station, enabling the method to well adapt to various complex working conditions and having stronger practicability and reliability. Brief Description of the Drawings

[0036] Figure 1 It is a schematic diagram of the step flow of the direct arc current monitoring method for the photovoltaic energy storage power station of the present invention. Detailed Embodiments

[0037] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. In addition, for convenience of description below, the "upper", "lower", "left", "right", etc. cited are in the same direction as the upper, lower, left, right, etc. of the accompanying drawings themselves. The "first", "second", etc. in the following text are for descriptive distinction and have no other special meanings.

[0038] The embodiment of the present application provides a method for monitoring the direct arc current of a photovoltaic energy storage power station, which solves the problems in the prior art. By parallelly extracting time-domain, frequency-domain and non-linear features and combining the improved D-S evidence theory for multi-feature fusion decision-making, the limitation of single-feature detection is effectively avoided, the false alarm rate is greatly reduced, and the detection accuracy is improved; multi-dimensional feature extraction can comprehensively capture the features of the current signal, enabling the system to accurately distinguish normal transient processes such as load switching and inverter start-stop from real arcs, avoiding false alarms and missed alarms, and ensuring the normal operation of the system; the dynamic threshold model enables the detection sensitivity to be automatically optimized with environmental changes, and the multi-feature fusion method fully considers the complex characteristics of the charging and discharging conditions of the photovoltaic energy storage power station, enabling this method to well adapt to various complex working conditions and having stronger practicability and reliability.

[0039] The technical solution in the embodiment of the present application to solve the above problems has the following general idea:

[0040] Embodiment

[0041] This embodiment gives a method for monitoring the direct arc current of a photovoltaic energy storage power station, as Figure 1 shown, including the following steps:

[0042] Step 1. High-frequency sampling: Obtain the DC-side current signal at a sampling rate of ≥100 kHz, and use a high-precision current sensor to collect the current signal to ensure the accuracy and stability of the collected data;

[0043] Among them, obtaining the DC-side current signal at a sampling rate of ≥100 kHz, the high sampling rate can more accurately capture the subtle changes of the current signal, provide a richer and more accurate data basis for subsequent feature extraction, and effectively avoid information loss caused by insufficient sampling rate;

[0044] Step 2. Multi-feature extraction: Calculate the time-domain feature kurtosis, impulse factor, current change rate (di / dt) respectively; Extract the energy distribution in the 3-100 kHz frequency band, and use wavelet packet decomposition to obtain the frequency-domain feature of the energy proportion of each sub-band; Calculate the non-linear features of sample entropy, Hurst exponent, and Lyapunov exponent;

[0045] Among them, time-domain features: Calculate kurtosis, impulse factor, current change rate (di / dt). Kurtosis is used to measure the degree of deviation of the signal from the normal distribution and can highlight the impact components in the signal; The impulse factor is sensitive to the impulse features in the signal and helps to detect the instantaneous impulse generated by the arc fault; The current change rate reflects how fast the current changes with time. When an arc fault occurs, the current change rate usually shows abnormal fluctuations;

[0046] Frequency domain features: Extract the energy distribution of the 3-100kHz frequency band, and use wavelet packet decomposition to obtain the energy proportion of each sub-band. This frequency band covers the main frequency components generated by arc faults. By analyzing the energy distribution and the energy proportion of each sub-band, we can more comprehensively understand the characteristics of the signal in different frequency bands and improve the accuracy of detection;

[0047] Nonlinear features: Calculate sample entropy, Hurst index, and Lyapunov index. These nonlinear features can mine complex nonlinear information in signals, effectively distinguish normal signals from arc fault signals, and enhance the detection capability of arc faults under complex working conditions.

[0048] Step 3: Dynamic threshold model: Based on the photovoltaic array operating voltage (v mp ), ambient temperature (T) to establish a threshold correction formula, which is derived based on a large amount of experimental data and theoretical analysis to achieve automatic optimization of detection sensitivity as the environment changes;

[0049] Among them, based on the photovoltaic array working voltage (v mp ), ambient temperature (T), and establish a threshold correction formula. The operating status of the photovoltaic power station is greatly affected by environmental factors. Through this dynamic threshold model, the detection sensitivity can be automatically optimized with environmental changes, ensuring that arc faults can be accurately detected under different environmental conditions;

[0050] Step 4: Improve DS evidence theory fusion: Introduce weighted credibility allocation, which initially allocates the importance of arc fault detection under different working conditions according to different features, and dynamically updates feature weights through Kalman filtering; Use Kalman filtering algorithm to dynamically update feature weights to adapt to different operating conditions, and trigger protection when fusion confidence ≥ 95%, with a response time of ≤ 30ms;

[0051] Among them, weighted credibility allocation is introduced, and the feature weights are dynamically updated through Kalman filtering. Taking into account the different contributions of different features to arc fault detection under different working conditions, the dynamic update of feature weights can more reasonably integrate multi-feature information and define decision rules: when the fusion confidence is ≥95%, the protection is triggered, and the response time is ≤30ms, ensuring that the requirements of rapid disconnection are met while accurately judging the arc fault, effectively reducing the risk of accidents.

[0052] The direct arc current monitoring method also includes recording the time of fault occurrence, current signal characteristics and environmental parameter information after the protection is triggered, so as to facilitate subsequent analysis of the cause of the fault.

[0053] During the entire monitoring process, the direct arc current monitoring method transmits the collected data, calculated characteristic values and monitoring status to the monitoring center through the communication module in real time to achieve remote monitoring and management.

[0054] The time-domain features include the kurtosis (K), pulse factor (PF), and current change rate (di / dt) of the current signal, where:

[0055]

[0056] μ is the current mean value, σ is the standard deviation, and I peak is the peak current.

[0057] In the dynamic threshold model, the threshold correction formula based on the operating voltage (v mp ) of the photovoltaic array and the ambient temperature (T):

[0058]

[0059] where I0 is the reference threshold, and α, β are correction coefficients.

[0060] The improved D-S evidence theory includes:

[0061] Assign initial credibility weights ω1, ω2, ω3 (Σwi = 1) to the time-domain, frequency-domain, and non-linear features respectively;

[0062] Dynamically update the weights through the Kalman filter, and the update formula is:

[0063] ω i (k) = ω i (k - 1) + K k *[z k - h(ω i (k - 1))]

[0064] K k is the Kalman gain, z k is the k-th observation value, and h(ω i (k - 1)) is the observation function.

[0065] By adopting the above technical solutions:

[0066] High-frequency sampling stage: Use a high-precision current sensor to collect the DC-side current signal of the photovoltaic power storage station at a sampling rate of ≥100 kHz to ensure the accuracy and stability of the sensor, so as to obtain accurate current signal data.

[0067] Multi-feature extraction stage:

[0068] Time-domain feature calculation: Use the collected current signal data to calculate the kurtosis, pulse factor, and current change rate (di / dt) through corresponding mathematical formulas.

[0069] Frequency-domain feature extraction: Perform Fourier transform on the current signal, extract the energy distribution in the frequency band of 3 - 100 kHz, and then use the wavelet packet decomposition algorithm to decompose the signal into different sub-bands and calculate the energy proportion of each sub-band.

[0070] Nonlinear feature calculation: Use a dedicated algorithm to calculate sample entropy, Hurst exponent, and Lyapunov exponent.

[0071] Dynamic threshold model establishment and application stage: Based on a large amount of experimental data and theoretical analysis, establish a threshold correction formula based on the working voltage (Vmp) and ambient temperature (T) of the photovoltaic array. During the actual operation process, real-time collect the working voltage and ambient temperature data of the photovoltaic array and substitute them into the formula to dynamically adjust the detection threshold.

[0072] Improved D-S evidence theory fusion stage:

[0073] Weighted credibility assignment: According to the importance of different features for arc fault detection under different working conditions, assign initial weighted credibility to each feature.

[0074] Kalman filter for dynamic update of feature weights: Use the Kalman filter algorithm, combined with the real-time collected data, to dynamically update the weights of each feature to make the weight assignment more reasonable.

[0075] Decision-making judgment: Compare the fused confidence with the set 95% threshold. When the fused confidence ≥ 95%, immediately trigger the protection mechanism and cut off the circuit within 30 ms to achieve a fast response to arc faults.

[0076] Finally, it should be noted that: Obviously, the above embodiments are only examples for clearly explaining the present invention, rather than limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to enumerate all the implementation manners here. And the obvious changes or modifications derived therefrom are still within the protection scope of the present invention.

Claims

1. A direct arc current monitoring method for a photovoltaic energy storage power station, characterized in that, It includes the following steps: Step 1, high-frequency sampling: Obtain the DC-side current signal at a sampling rate of ≥100 kHz; Step 2, multi-feature extraction: Calculate the time-domain feature kurtosis, pulse factor, and current change rate (di / dt) respectively; Extract the energy distribution in the 3-100 kHz frequency band, and use wavelet packet decomposition to obtain the frequency-domain feature of the energy ratio of each sub-band; Calculate the non-linear features of sample entropy, Hurst exponent, and Lyapunov exponent; Step 3. Dynamic threshold model: Based on the working voltage (v mp ) of the photovoltaic array and the ambient temperature (T), establish a threshold correction formula; Step 4, improved D-S evidence theory fusion: Introduce weighted credibility assignment, and dynamically update the feature weights through Kalman filtering; When the fusion confidence level ≥95%, trigger protection, and the response time ≤30 ms.

2. The direct arc current monitoring method for a photovoltaic and energy storage power station according to claim 1, characterized in that In the high-frequency sampling step in Step 1, a high-precision current sensor is used for current signal acquisition to ensure the accuracy and stability of the acquired data.

3. The direct arc current monitoring method for a photovoltaic and energy storage power station according to claim 1, characterized in that In the dynamic threshold model in Step 3, the threshold correction formula is obtained based on a large amount of experimental data and theoretical analysis to realize the automatic optimization of the detection sensitivity with the change of the environment.

4. The direct arc current monitoring method for a photovoltaic and energy storage power station according to claim 1, characterized in that In the improved D-S evidence theory fusion in Step 4, the weighted credibility assignment is initially assigned according to the importance of different features for arc fault detection under different working conditions.

5. The direct arc current monitoring method for a photovoltaic and energy storage power station according to claim 1, characterized in that In the improved D-S evidence theory fusion in Step 4, the Kalman filter algorithm is used to dynamically update the feature weights to adapt to different operating conditions.

6. The direct arc current monitoring method for a photovoltaic and energy storage power station according to claim 1, wherein The direct arc current monitoring method also includes recording the time of fault occurrence, current signal characteristics, and environmental parameter information after triggering protection for subsequent analysis of the fault cause.

7. The direct arc current monitoring method for a photovoltaic and energy storage power station according to claim 1, characterized in that During the entire monitoring process, the direct arc current monitoring method transmits the acquired data, calculated feature values, and monitoring status to the monitoring center through the communication module in real time to achieve remote monitoring and management.

8. The direct-arc current monitoring method for a photovoltaic and energy storage power station according to claim 1, wherein The time-domain features include the kurtosis (K), pulse factor (PF), and current change rate (di / dt) of the current signal, where: μ is the current mean, σ is the standard deviation, and I peak is the peak current.

9. The direct arc current monitoring method for a photovoltaic and energy storage power station according to claim 1, characterized in that In the dynamic threshold model, the threshold correction formula based on the operating voltage (v mp ) of the photovoltaic array and the ambient temperature (T): where I0 is the reference threshold, and α, β are correction coefficients.

10. The direct arc current monitoring method for a photovoltaic and energy storage power station according to claim 1, characterized in that, The improved D-S evidence theory includes: Initial credibility weights ω1, ω2, ω3 (Σwi = 1) are respectively assigned to the time-domain, frequency-domain, and non-linear features; Dynamically update the weights through the Kalman filter, and the update formula is: ω i ω(k) = i ω(k - 1)+K k *[z k - h(ω i (k - 1))] K k is the Kalman gain, z k is the k-th observation value, h(ω i (k - 1)) is the observation function.

Images