MAIRS is an online “artificial intelligence†service in the data analysis service provided by Wisdom, and is an extension service of the IOTAS industrial IoT data analysis service system INDASS, which provides continuous identification risk identification for industrial equipment.
The so-called risk identification, that is, the possibility of determining the risk in the future time period, including: the risk of equipment failure, the operational stability of the equipment (or various problems caused by stability), and the equipment There is a risk of a state mutation.
MAIRS is actually a recognition model for industrial data, a series of calculation methods for industrial data. We simply understand MAIRS as a robot that can identify industrial data.
When MAIRS was born, although there was a brain, it didn't have a brain because MAIRS had no knowledge. So, we have to cultivate it first. So, we have to do two things with MAIRS:
The first one is deep learning, which is to use the existing data samples to train the model, which is to let MAIRS learn "knowledge";
The second is to turn the learned "knowledge" into a service that identifies the condition of the operation of the equipment and judges various risks in the future.
Let me talk about the training of MAIRS. The so-called training is to use the data that has been collected from many devices before to make a "training sample", and use this sample to continuously train this MAIRS model. This training can be described as "the strength of the family", because MAIRS is to learn from all similar devices, so that even in the face of a new online device, it can predict that this device has never appeared. At the same time, MAIRS's learning and training is a process of "live to learn old". Uninterrupted learning, MAIRS's knowledge is constantly enriched, and the accuracy of recognition continues to increase.
Let's focus on what MAIRS recognizes.
Remember the INDASS, the real-time data of the device is indexed in the INDASS. In addition to the level of the index, a three-dimensional gradient calculation is performed on the index in MAIRS. In other words, in addition to understanding the current level of the index, we need to know how this index has changed. Just like the stock index, although it is 3,000 points, the drop from 3,500 points is still up from 2,500 points, and the meaning is completely different. In addition to the exponential and exponential gradients, what is left is the characteristic value of the device, what temperature, pressure, current and voltage.
The first thing to understand is that MAIRS's data source is INDASS, which is the data analysis (computing) service of Wisdom. So, what is the result of MAIRS?
What we see is that our recognition is divided into several parts:
(1) Comprehensive assessment. The front is the calculated number, which is a combined value, representing a series of possible situations; followed by the identification, such as N, OQ, OA, OF, etc., which represent possible situations:
• N stands for Normal and everything is ok;
• OQ stands for Over Quota, which means that data may be out of range;
• OE stands for On Event, meaning that some unexpected events may have occurred;
• OF stands for On Fault, meaning that some faults may have occurred;
• OA stands for On Alert, meaning that some alerts may have occurred.
(2) Probability of failure, stability, and mutation risk in four prediction intervals (10, 30, 60, and 90 minutes) (0.00 to 1.00, that is, 0% to 100%). It should be noted here that this prediction is based on the prediction of "continuous operation", which means that the device will not start and stop in the middle.
The MAIRS model is specifically designed for industrial equipment operation identification. It is different from other AI (artificial intelligence) applications. It is not as simple as "learning tens of thousands of cat photos and then identifying a cat". First of all, MAIRS is a dynamic recognition model. The reason is very simple, because only the current equipment parameters, temperature and pressure, current and voltage frequency, not only can not judge what will happen in the future, even if it is currently what state is difficult to judge.
That's why, in addition to identifying the level of the index, we need to identify the gradient of the exponential change. The exponential gradient is actually a mathematical concept, meaning the direction of the exponential change and the degree of change. In other words, the basis for MAIRS identification is not only the level of the index, but also the change in the index.
Let us give an example: I assume that we judge the stock market situation based on the current Shanghai Stock Exchange Index. In fact, we can't know anything from the current 3000 points. However, if we know that today's 3000 points are up from 2,500 points, then we can basically judge that the stock market is getting better, if this is three days. The thing is that the stock market is skyrocketing. If it is a three-month event, it is only getting better. Similarly, if we know that today’s 3000 points fell from 3,500 points three days ago, it will definitely judge the stock market. In a plunge.
In the same prediction interval, three things are actually predicted: fault, stability, and mutation.
(1) Failure, this easy to understand is the possibility of failure in this prediction interval.
(2) Stability, meaning that in the prediction interval, the device has stability problems, or the possibility of other device problems due to stability problems. We should still remember the calculation of the stability of the equipment parameters in INDASS. Here, it is actually based on the calculation of the stability of the important parameters of INDASS, and the overall stability trend judgment of the stability of the equipment.
(3) Mutation, that is, a sudden change in state, for example, a device crash is a mutation. To give an extreme example, a device with a rated pressure of 10 MPa, when it reaches 15 MPa, exceeds the safety threshold, but once it reaches 20 MPa, an explosion may occur, which is a sudden change. MAIRS fits according to the mutated folding model and predicts the risk of mutation in the interval for continuous operation of the device.
MAIRS is a new member of the IoT Industrial IoT system. In the near future, MAIRS will become the patron saint of the operation of the equipment.
Solder is an indispensable material for the manufacture of electronic products
Solder Wire,Lead Free Solder Wire,Solid Solder Welding Wire,Solder Tin Wire
Shaoxing Tianlong Tin Materials Co.,Ltd. , https://www.tianlongspray.com