Traditional oil analysis, also called oil condition monitoring (OCM) analyzes samples for abnormal wear, condition of the fluid, and presence of contaminates. The presence of any of these issues indicates that action needs to be taken to correct the problem in order to maintain the efficiency, effectiveness, and lifespan of the equipment. The disadvantages of the traditional method of oil analysis are that the process occurs after the problems have happened and is limited due to human and time constraints.
To start the process companies send samples to a laboratory for analysis. Laboratories receive millions of such samples each year, looking at 100 data points per sample. Unfortunately looking at only 100 data points is not truly accurate, but due to the sheer number of samples being sent in for testing, labs are unable to test further. Analyzing one sample takes up to five minutes, so the time need to analyze millions of samples is prohibitive.
The other issue related to traditional oil analysis is that it takes place after the problems have occurred. Consequently, the lifespan and efficiency of the machinery has already been negatively affected by the time the engineers receive the results. This, in turn, negatively influences the revenue stream of the company.
Looking for methods to improve oil analysis has led researchers to explore the concepts of incorporating artificial intelligence and machine learning. Artificial intelligence (AI) allows machines to learn from experiences in order to adjust to new inputs. Machine learning is a branch of artificial intelligence. It is based on the concept that systems can learn from inputted data, identify patterns from that data, and learn to make decisions with little human intervention. Essentially, it is possible to train computers to accomplish specific jobs by processing copious amounts of data and recognizing patterns that exist within the data.
One clear advantage of AI and machine learning over traditional analysis is its ability to look at thousands of data points, in a shorter period of time. The analysis is more in-depth, providing analysts and companies with a fount of information to increase the efficiency and effectiveness of equipment, increase revenue streams, and decrease potential equipment down time
In addition, as the machine learning continues its analysis it is able to identify indicators of equipment wear or fluid condition changes. Companies can then take action before the issue becomes a problem, and increase the lifespan and efficiency levels of the machinery. This is possible because the information input for the machine learning process to occur is divided into two separate parts. The first part provides the information necessary for the actual machine learning to take place. The second part is used to test the system, during which time the model looks for any identifying features or combinations of features from all the information provided. All features and combinations of features identified are then used when analyzing future samples, looking for those same indicators of issues within the equipment or fluids.
Artificial intelligence and machine learning can clearly revolutionize the field of oil analysis, resulting in faster, more in-depth analysis, and providing solutions before issues turn into big problems. The results could mean large savings for industries, companies, and consumers themselves.