A photovoltaic DC series arc detection method based on dynamic reference

Abstract

A photovoltaic DC arc detection method based on a dynamic benchmark includes the following steps: current sampling to obtain time-series current data; calculating the current fluctuation amplitude, calculating a current fluctuation amplitude benchmark, and updating a current fluctuation amplitude counter by comparing the current fluctuation amplitude with the current fluctuation amplitude benchmark; calculating the high-frequency component ratio, calculating a high-frequency component ratio benchmark, and updating a high-frequency component ratio counter by comparing the high-frequency component ratio with the high-frequency component ratio benchmark; and determining whether a DC arc has occurred in the current time period based on the counts of the current fluctuation amplitude counter and the high-frequency component ratio counter. This method does not require high-performance real-time computing support, is relatively low-cost, and is suitable for low-cost scenarios.

Description

Technical Field

[0001] The present invention belongs to the field of arc detection, and in particular relates to a photovoltaic DC series arc detection method based on a dynamic reference. Background Art

[0002] Currently, many photovoltaic installations are experiencing the highly dangerous DC series arcing caused by poor construction and installation quality or loose connections. Since DC arcs lack a zero-extinguishing characteristic, they continue to burn, causing significant damage and a high risk of fire.

[0003] Existing DC arc detection methods are mostly based on the high-frequency characteristics of arc generation, wavelet transforms, and the rate of change of current fluctuations. However, most methods ultimately rely heavily on the selection of thresholds. For example, characteristic values calculated using the various methods mentioned above are compared with fixed thresholds to determine whether DC series current is occurring. Fixed-threshold methods are firstly sensitive to the threshold; even the slightest inappropriate threshold selection can significantly increase the false alarm rate or missed alarm rate. Secondly, as the environment and load vary, it is almost impossible to use a single threshold for all scenarios. An algorithm that performs well in one scenario may perform poorly depending on the scenario. Summary of the Invention

[0004] The object of the present invention is to provide a photovoltaic DC series arc detection method based on a dynamic reference to solve the problems raised in the background technology.

[0005] A photovoltaic DC arc detection method based on a dynamic benchmark comprises the following steps: current sampling to obtain current data based on a time series; calculating the current fluctuation amplitude, calculating the current fluctuation amplitude benchmark, and updating the current fluctuation amplitude counter by comparing the current fluctuation amplitude with the current fluctuation amplitude benchmark value; calculating the high-frequency component ratio, calculating the high-frequency component ratio benchmark, and updating the high-frequency component ratio counter by comparing the high-frequency component ratio with the high-frequency component ratio benchmark value; and determining whether a DC arc occurs in the current time period based on the counts of the current fluctuation amplitude counter and the high-frequency component ratio counter.

[0006] Preferably, the acquisition of time-series-based current data is to acquire a current sampling sequence at any moment using 4096 sampling points as a window, calculate the average value of the current sampling sequence as the current value at the corresponding moment, and obtain a current sequence that changes with time.

[0007] Preferably, the current fluctuation amplitude calculation includes the following steps: dividing the current sequence into several current groups according to the preset time length, rearranging the elements in any current group according to the size; for any rearranged current group, calculating the corresponding current fluctuation amplitude Among them, IPi represents the i-th current value in the rearranged current group, α, β, and θ are preset values, α>0, β+θ<n, and n represents the number of elements in the current group.

[0008] Preferably, the calculation formula of the current fluctuation amplitude reference is:

[0009]

[0010] in, is the current fluctuation amplitude benchmark of the current group, is the current fluctuation amplitude benchmark of the previous current group, P update is the iterative update ratio of the current fluctuation amplitude benchmark, Fr s is the current fluctuation amplitude.

[0011] Preferably, the current fluctuation amplitude counter is updated by comparing the current fluctuation amplitude with the current fluctuation amplitude reference value as follows:

[0012]

[0013] Among them, Counter Fr is the count of the current fluctuation amplitude counter, Fr s is the current fluctuation amplitude, It is the current fluctuation amplitude benchmark.

[0014] Preferably, the high-frequency component ratio calculation includes the following steps:

[0015] Perform Fourier transform on the current sampling sequence at any moment to obtain the high-frequency component;

[0016] Calculates the high frequency components and Mag within the preset frequency range high_sum , and calculate the proportion of high-frequency components Mag dc is the DC component.

[0017] Preferably, the preset frequency range is 40KHz-100KHz.

[0018] Preferably, the calculation formula of the high-frequency component ratio benchmark is:

[0019]

[0020] in, is the current high-frequency component ratio benchmark, is the previous high-frequency component ratio benchmark, W z is the current high frequency component ratio, Q update Iterative update ratio of the high-frequency component ratio benchmark.

[0021] Preferably, the high frequency component ratio counter is updated by comparing the high frequency component ratio with the high frequency component ratio reference value as:

[0022]

[0023] Among them, Counter W is the count of the high frequency component proportional counter, is the current high-frequency component ratio reference value, W z is the current high-frequency component ratio.

[0024] Preferably, the determination of whether a DC arc occurs based on the counts of the current fluctuation amplitude counter and the high-frequency component proportional counter is: when the count of the current fluctuation amplitude counter is greater than 2 and the count of the high-frequency component proportional counter is greater than 2, a DC series arc occurs, otherwise it does not occur.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] The present invention determines whether a DC arc occurs by comparing the changes between the current calculated value and the reference value. There is no need to manually set any thresholds according to the scenario, and the reference value is dynamically updated. It can self-iterate and adapt to the new environment after a certain period based on the current weather, temperature, and load. At the same time, this method does not require real-time computing support with ultra-large computing power, has low cost, and is suitable for low-cost scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 Flowchart of the present invention. DETAILED DESCRIPTION

[0028] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0029] A photovoltaic DC arc detection method based on dynamic reference, the method specifically includes steps.

[0030] Step 1: Current sampling, obtain the current sampling sequence, and calculate the current data based on the time series.

[0031] Specifically, the current sampling sequence I = {I1, I2, ..., I 4096}, calculate the average value of the current sampling sequence i represents the sampling point number. The average value of the current sampling sequence is used as the current value at the corresponding moment, resulting in the time-varying current sequence {Ip1, Ip2, Ip3, ...}. Ip1 represents the average value of the current sampling sequence at the first moment, that is, the current value at the first moment.

[0032] In step 1 of the present invention, in order to eliminate normal sampling fluctuations and interference, the average value is calculated using 4096 sampling points as a window, which serves as the basis for calculating the current fluctuation amplitude in step 2.

[0033] Step 2, calculate the current fluctuation amplitude, calculate the current fluctuation amplitude reference, and update the current fluctuation amplitude counter by comparing the current fluctuation amplitude with the current fluctuation amplitude reference value. Specifically, it includes 4 sub-steps.

[0034] Step 2.1, set the preset time length to 30 time moments, divide the current sequence into several current groups according to the preset time length, and rearrange the current values in any current group from large to small {Ip1, Ip2, ..., Ip 30};

[0035] Step 2.2: For any rearranged current group, calculate the corresponding current fluctuation amplitude Among them, IP j represents the jth current value in the rearranged current group.

[0036] Step 2.3, calculate the current fluctuation amplitude reference based on the current fluctuation amplitude. The calculation formula of the current fluctuation amplitude reference is:

[0037]

[0038] Among them, P update It is the iterative update ratio of the current fluctuation amplitude benchmark, usually 0.1-0.2, is the current fluctuation amplitude benchmark of the current group, It is the current fluctuation amplitude benchmark of the previous current group.

[0039] In the present invention, the initialization current fluctuation amplitude reference value is 0, that is, The current fluctuation amplitude reference is dynamic, and there is a corresponding current fluctuation amplitude reference for each current fluctuation amplitude.

[0040] Step 2.4, update the current fluctuation amplitude counter based on the following formula,

[0041]

[0042] Among them, Counter Fr is the count of the current fluctuation amplitude counter, Fr sis the current fluctuation amplitude, It is the current fluctuation amplitude benchmark.

[0043] Step 3, calculate the high frequency component ratio, calculate the high frequency component ratio benchmark, and update the high frequency component ratio counter by comparing the high frequency component ratio with the high frequency component ratio benchmark value, which specifically includes 4 sub-steps.

[0044] Step 3.1, the current sampling sequence I={I1,I2,...,I 4096}Perform Fourier transform to obtain high-frequency components.

[0045] Step 3.2, calculate the high frequency components and Mag within the preset frequency range of 40KHz-100KHz high_sum , and calculate the proportion of high-frequency components Mag dc is the DC component.

[0046] In the present invention, the high frequency of a photovoltaic DC fault arc is usually between 40 and 100 kHz. According to the Nyquist sampling theorem, this solution adopts a current sampling frequency of 200 kHz to ensure the retention of the photovoltaic DC arc signal.

[0047] Step 3.3: Calculate the high-frequency component ratio benchmark W based on the high-frequency component ratio base , the calculation formula of the high-frequency component ratio benchmark is

[0048]

[0049] Among them, Q update It is the iterative update ratio of the high-frequency component ratio benchmark, which is generally 0.1-0.2; is the current high-frequency component ratio benchmark, is the previous high-frequency component ratio benchmark, W z is the current high-frequency component ratio.

[0050] In the present invention, the high frequency component ratio reference value is initialized to 0, that is, The high-frequency component ratio benchmark is dynamic, and each high-frequency component ratio has a corresponding high-frequency component ratio benchmark.

[0051] Step 3.4, update the high frequency component proportional counter according to the following formula:

[0052]

[0053] Among them, Counter W is the count of the high frequency component proportional counter, is the current high-frequency component ratio reference value, W z is the current high-frequency component ratio.

[0054] Step 4: Determine whether a DC arc occurs in the current period based on the counts of the current fluctuation amplitude counter and the high-frequency component proportional counter.

[0055] Specifically, when the count of the current fluctuation amplitude counter is greater than 2 and the count of the high-frequency component proportional counter is greater than 2, the DC series arc occurs, otherwise it does not occur.

[0056] In the present invention, after a period of iteration, Fr base and W base Will automatically adapt to the current environment.

[0057] In the present invention, it should be noted that all constant parameters appearing in the above steps can be set as empirical values, but there are no strict requirements for parameter selection. A certain deviation will not affect the judgment result, and there is no need to adapt according to the environment.

Claims

1. A photovoltaic DC arc detection method based on dynamic reference, characterized in that: The method comprises the following steps: Current sampling, obtaining current data based on time series; Calculating the current fluctuation amplitude, calculating the current fluctuation amplitude reference, and updating the current fluctuation amplitude counter by comparing the current fluctuation amplitude with the current fluctuation amplitude reference value; The current fluctuation amplitude calculation comprises the following steps: Divide the current sequence into several current groups according to the preset duration, and rearrange the elements in any current group according to their size; For any rearranged current group, calculate the corresponding current fluctuation amplitude , among which, IP j represents the jth current value in the rearranged current group, α, β, and θ are preset values, α>0, β+θ<n, and n represents the number of elements in the current group; The calculation formula of the current fluctuation amplitude reference is: ， in, is the current fluctuation amplitude benchmark of the current group, is the current fluctuation amplitude benchmark of the previous current group, P update is the iterative update ratio of the current fluctuation amplitude benchmark, is the current fluctuation amplitude; The current fluctuation amplitude counter is updated by comparing the current fluctuation amplitude with the current fluctuation amplitude reference value. ， in, is the count of the current fluctuation amplitude counter, is the current fluctuation amplitude, It is the current current fluctuation amplitude benchmark; Calculating the high frequency component ratio, calculating the high frequency component ratio reference, and updating the high frequency component ratio counter by comparing the high frequency component ratio with the high frequency component ratio reference value; The high frequency component ratio calculation comprises the following steps: Perform Fourier transform on the current sampling sequence at any moment to obtain the high-frequency component; Calculates the high frequency components and Mag within the preset frequency range high_sum , and calculate the proportion of high-frequency components , Mag dc is the DC component; The calculation formula of the high-frequency component ratio benchmark is: ， in, is the current high-frequency component ratio benchmark, is the previous high-frequency component ratio benchmark, is the current high-frequency component ratio, is the iterative update ratio of the high-frequency component ratio benchmark; The high-frequency component ratio counter is updated by comparing the high-frequency component ratio with the high-frequency component ratio reference value, which is expressed as: ， in, is the count of the high frequency component proportional counter, is the current high-frequency component ratio reference value, is the current high-frequency component ratio; Whether a DC arc occurs in the current period is determined based on the counts of the current fluctuation amplitude counter and the high-frequency component proportional counter.

2. A photovoltaic DC arc detection method based on dynamic reference according to claim 1, characterized in that: The current data based on time series is obtained as follows: The current sampling sequence at any moment is obtained using 4096 sampling points as a window, and the average value of the current sampling sequence is calculated as the current value at the corresponding moment to obtain the current sequence that changes with time.

3. The photovoltaic DC arc detection method based on dynamic reference according to claim 1, characterized in that: The preset frequency range is 40KHz-100KHz.

4. A photovoltaic DC arc detection method based on dynamic reference according to any one of claims 1 to 3, characterized in that: The method for determining whether a DC arc occurs based on the counting of the current fluctuation amplitude counter and the high-frequency component proportional counter is as follows: When the count of the current fluctuation amplitude counter is greater than 2 and the count of the high-frequency component proportional counter is greater than 2, a DC series arc occurs, otherwise it does not occur.

Images