At 10:49:20AM 2022-10-17 - BSM2_model_old_flowcombiner works fine when Qr = 0.
However, there is error when Qr flow = 20,648 is added to Activatedsludge input port.
Asking Brian why there was no error when running the model last Friday.
Answer from Brain: Something is causing Proc7 to be NaN - not a number - most likely caused by a division by zero. The sim attempts to continue after that occurs around sim time of 0.26 but really everything after that is bad data. There was a change to limit the number of times this message could show up in a simulation before shutting down to 10,000. That could be the difference between Friday and Today. The 10,000 limit just prevents simulations from ending up in an infinite loop.
He let me know that I am setting this 0 in the x_gt_or_equai_zero py script within the temp_ODE_X submodel and suggested me to use another value that is close to zero.
My reply to him: I am going to set the Q_r flow to zero from now. I think the problem is when I set Q_r to be 20,648 as a placeholder for my return flow from the settler, I still have other variables, such as the concentration of soluble materials and particulate materials of the return flow, to be zero. As a result, the concentration of the flow got diluted over time and eventually, the concentration of particulate material was equal to zero. I will follow your suggestion of changing the value in my x_gt_or_equal_zero py script if I still encounter the same error once I finish creating my settler block.
update all the question that needed to get answer from JoaKim
histogram of water level distribution
write an one page of how I am going to use all_frames_SANMN00893_1.csv to label all the DNV dataset. Once we get the label, we can use the data to train the model and to use image to evaluate the performance of SewelML on DNV dataset.
create a list of error when generating all_frames_SANMN00893_1.csv such as why some certain images did not have labels even though there was observation text.
Assess dataset: need to find out if LOF of pipe reported in Assess dataset was done by segment of sewer main or the whole sewer main. If yes, we need to find a way to calculate LOF of the pipe by using the LOF of each pipe's segment.
Assign COF for each factor according to Vladenu et al 2019 title "Consequence-of-Failure Model for Risk-Based Asset Management of Wastewater pipes Using AHP"
Share the Vladeanu paper and Saaty,1980 with Dr.Lence. We are going to discuss about this on wednesday
Need to rewrite the proposed step to get the defect label
search on COF using in Scream. Read the paper about scream.
Email the author if SCREAM is doing with COF
Get the csv for keywords (there is problem with getting the keywords in the articles)
Submit the job to compute canada
NPL: I think Vannary needs to document the findings and keep bugging the support. Vannary, please put together a document for support people and cc me when you send it out.
set aside 5 videos randomly for 10 combination or so. Then train the model on the rest of video and test on the test dataset. If all the models are consistent, see how well. Do draw how I grab the image for training testing
Make a histogram of defect of dataset. Decide whether we want balance the dataset or not.