As the most commonly used equipment in photovoltaic power plants, combiner boxes are often prone to various faults. Operation and maintenance personnel need to find out the fault points through inspections. Taking a 20MW power station as an example, in the traditional purely manual inspection method, it basically takes 1 month to inspect the combiner box in the station once. The inspection of the appearance occupies most of the workload, and the inspection of the internal combiner box is a random inspection method, which is difficult to investigate in depth. The only way to find the fault is the current value of the station control system, but the actual situation is that the attendant only checks the inverter For the power generation status of the combiner box, the data volume at the combiner box level is large and unintuitive, and it is impossible to locate and trace the faulty equipment, which provides great difficulty for the supervisors. The big data solution of the SANAS management and control platform As the first monitor for the hidden troubles and troubleshooting of the combiner box, the solution is gradually gaining popularity.
Early warning of combiner box
The big data monitoring program is to conduct daily diagnosis for each combiner box of the whole station, give the daily fault diagnosis results of each combiner box, and understand the operating status of the combiner box-level equipment at a glance, and guide the personnel in the station to conduct targeted inspections. Inspection and troubleshooting, supervising the personnel in the station to restore the faulty equipment, the discovery and repair of the fault has been substantially improved, so that minor faults are no longer in a state of no one to care about. According to the results of the power station calculation, the monitoring method of the big data solution can improve the timeliness of fault detection by 90% and shorten the time of faults by 70% compared with the traditional inspection method.
Efficiency improvement point: Compare the operation status of the combiner box, check the power generation status of the string for the early warning equipment, and determine the location of the faulty string.
Steps to improve efficiency
The first step: data cleaning
Set a certain branch current threshold, remove data higher or lower than this value in the communication process, and remove abnormal data and the value that is the same as the branch current at the adjacent time. The purpose is to reduce the large negative value generated during the communication process. And Dazheng value is screened out, and valuable data is retained for the next step.
The second step: branch selection
According to the actual number of equipment connected to the branch string, select the actual number of string current values, calculate the sum of the current values of each branch for a period of time, sort from small to large, take the first 12 (the combiner box returns 16 each time The branch current value, but only 12 channels are actually connected) The branch with the largest sum of string current values is used as the branch of the actual connected device, and the standard deviation between the branch current values at each time is calculated and filtered again.
The third step: Calculate the weighted dispersion rate
The data of each branch and each time obtained through the above screening, calculate the standard deviation and average value of the branch current at each time, and use the average value of the branch current at each time as the weight to calculate the weighted average of the branch current variation coefficient at each time value;
The fourth step: Early warning judgment.
Through the above calculation and screening, a constant value can be obtained. By comparing the branch current value of the combiner box with it, it can be judged whether the combiner box is faulty.