Niche topic: predicting and preventing catastrophic events from the perspective of event probability
IT Little Stonemason Lao Chang
Transferred from: WeChat official account, IT small stone mason Lao Chang
Didn't it say yesterday that there would be a rainstorm this morning? What about the rainstorm? Come out and see if I can strangle you. Just at the end of the article, rainstorm said in a hurry: "I have something to do on the way, and I was delayed. Now I have arrived in Tongzhou and Daxing, and I will soon arrive at Chaoyang. Please wait patiently for the people in Chaoyang. Sorry, sorry, I'm late. I hereby solemnly promise that I will stay in Chaoyang tonight and will not leave"!
It is said that the prediction of small probability events is much more difficult than the prediction of large probability events. In the north in summer, the success rate of predicting sunny days is much higher than that of predicting rain. There is a big head to chew here. First, there are fewer small probability events such as rain and less data accumulation. The weather forecast model based on data validation may be less reliable than the fire weather. Secondly, the probability of small probability events occurring is low, and coupled with inaccurate predictions, the psychological impression given to the public is likely to be that the accuracy of predictions is relatively low. Friends from the south, can you also count or pay attention to the fact that summer in the south is a high probability event with heavy rain, so will your meteorological bureau in Gada have a higher success rate in predicting rainy days?
The relationship between predicted success rate and target probability.
1、 Theoretically equal probability, but in reality it may not necessarily be the case
There are two bridges in Beijing that seem to be against me. One is the Guomao Bridge, the other is the Gengzhuang Bridge, the other is in the place where I used to work, and the other is near where I live; The impression statistics from many years in the past indicate that under the Guomao Bridge, from west to east; Gengzhuang Bridge runs from south to north; Being stuck is an absolute high probability, eight out of ten. I am very sensitive and curious about things like this. I have tried to explore various reasons, but it seems that none of them can convince me. In the end, it can only be understood as the sample size is still too small and the time traversal is not long enough.
2、 Temporal and spatial traversal of probability
We still need to return to theory, which states that as long as there are enough spatial and temporal traversals, the probability events should be equal. For example, when buying a lucky lottery ticket, if you buy the same bet every time, the probability of winning the first prize is approximately 97000 years, which is time travel; If you spend 35 million yuan to purchase all the combinations at once, tomorrow night (Thursday) you will receive all the awards, including the first prize, which is called space traversal. This case demonstrates that spatial traversal is the most effective if possible, as it can save 97000 years of long waiting. This is what we see. Many times, there are many things that require a lot of manpower.
3、 Gambling under Probability
No matter how many people can realize the real practical significance of probability, gambling is still one of the most exciting activities in the world. Las Vegas in the United States and Macao in China are the most famous gambling cities in the world. Wu Yang, a gambler, flocks to gambling cities, hoping to obtain the pleasure that is hard to reach in limited time and space through the use of qi (contingency).
Jade gambling stones also carry this risk. I once wrote an article titled "Jade Gambling Stones: Small Gambling Enjoys Emotions, Big Gambling Hurts the Body" (302 readings, available on Baidu). The risk of gambling stones also comes from counterfeiting, which adds insult to injury to injury for those who do not understand gambling stones.
In fact, compared to gambling with almost infinite sample space and boundless sky like lottery, for experienced people, there is at least a 30% probability that jade gambling stones are positively correlated with experience. I am quite an experienced gambling player. If someone's hand itches, I have a prescription for relieving itching for you.
4、 What are the reasons for the uncertainty of probability
Apart from the Guomao Bridge and Gengzhuang Bridge mentioned earlier, there seem to be three factors that I can understand that lead to uncertainty in probability:
1. Determined at the micro level, but uncertain at the macro level
There should be many examples in this regard, such as how many people do we predict to visit the Forbidden City tomorrow? From a micro perspective, the load forecasting of the power grid, and even the demand side management of electricity, are basically planned by these individuals. However, we cannot or find it difficult to master these individual plans, which leads to macro uncertainty. Macro uncertainty requires the use of mathematical statistical methods, whether based on probability, fuzzy prediction, or other mathematical models.
Nowadays, many people who engage in big data and artificial intelligence in the field of electricity do want to use methods such as metering and billing to grasp the electricity consumption of all users (residents, enterprises, etc.), in order to achieve the best so-called accurate load forecasting. I would like to say that the electricity consumption information obtained through these methods indicates the past and present, and cannot represent the future, but can only serve as a reference for future forecasting. For example, the Lao Li family usually has three people, and their parents come from the countryside to live for a while. However, the goose will definitely increase the electricity consumption of the Lao Li family in the future.
Load forecasting should be classified, such as commercial electricity, electricity used by government agencies and institutions, electricity used by enterprises, and electricity used by residents, and then integrated. Among them, commercial electricity, electricity used by government agencies and institutions, and electricity used by residents have strong macro regularity and are highly related to factors such as temperature changes, commuting hours, and holidays; Predicting is not difficult, as both residential electricity meters and enterprise gateway meters can obtain these data. The uncontrollable factor in the future may be electric vehicle charging.
From a micro perspective, the electricity consumption of enterprises should also be manageable, as they have production plans that are highly correlated with the electricity and energy consumption of production equipment. Compared to residents, the number of enterprises is still relatively small, and it is still possible to obtain these data by finding ways. The only trouble lies in how to determine the manufacturing and processing enterprises in the secondary industry, The cross checking relationship between their production plans (different products and different process routes) and the energy consumption of the enterprise means that a company can calculate the expected consumption of water, electricity, gas, and heat based on different products and production processes from their next week's production plan, and this estimate should be relatively accurate. Sometimes it's a bit confusing for companies that provide cloud computing services. They show off their so-called cloud monitoring platforms all day, how well they are doing and how colorful they are. It's better to sink in and create an Addon in the ERP software of manufacturing enterprises. When the ERP plan is launched, it's more valuable to calculate the relevant energy consumption with it. In the past two years, when I was doing power EPC, I often had such thoughts and impulses, but it seems that no one understands this matter. The second is how to obtain these data, which is within the scope of demand side management. In my understanding, demand side management is a social movement, a big mechanism and synergy for achieving technology and benefit sharing in a broader field. It's not just about the power grid company.
The data from the past year shows that the proportion of electricity consumption in the secondary industry has exceeded 70% of the total electricity consumption in society. The saying that we should focus on the main contradiction in our work is probably the same.
For example, being regarded as the two largest and blackest black swans in the world: firstly, Brexit; Secondly, Trump's election is actually a pseudo random event, but many people are not aware that global economic uncertainty leads to political uncertainty and the proliferation of Western populism. What will be the result of the combined fermentation of these social factors?
2. Uncontrollable causal relationships
I personally believe that the problem mentioned earlier should be pseudo random; But the random probability problem with uncontrollable causal relationships should be true randomness. Completely different from the 30000 people preparing to go to the Forbidden City tomorrow, of which 15000 have already applied on the Forbidden City website, and at least 10000 people will apply on the website in the afternoon or go to the scene to buy tickets tomorrow, which is a pseudo random process. The lucky lottery you bought today (Wednesday) will only be known when the lottery is opened tomorrow night. When the lottery is opened, it is the result of many balls constantly colliding, and this causal relationship that causes randomness changes dramatically during the random process, resulting in unpredictable or extremely difficult to predict situations, which can be considered as true randomness.
3. Impact of unexpected events
Sudden events can be divided into three categories: natural disasters, man-made disasters, and huge secondary disasters caused by natural or man-made disasters. There should be many natural and secondary disasters that everyone is familiar with or has experienced, such as meteorite impacts, earthquakes, floods, epidemics, ice disasters, mountain fires, and secondary disasters such as mudslides, landslides, and tsunamis.
3.1 Natural disasters gradually move from unpredictability to predictability and prevention
In the past, many natural disasters were unpredictable because they were basically sudden. When they arrived, the only methods that weak humans could take were to hide, hide, heave a long sigh, and helplessly let them ravage.
It is said that the dinosaurs that dominated the Earth at that time were severely impacted by a huge meteorite 65 million years ago, resulting in decades of starvation or burning to death due to the obscuration of the sun. Geologists discovered the deposition of metallic iridium in the same era's strata.
But there are also different schools of thought that the Indian subcontinent "plans" to meet the Eurasian plate, separate from a super ancient continent called Gondwana, and along the way to the north, a place called Deccan had a long-term, large-scale volcanic eruption. For this reason, I also wrote an article titled "It's not easy to figure out who ended the wonderful life of dinosaurs 65 million years ago" (189 reads, available on Baidu).
However, with the enhancement of human technological capabilities and the continuous deepening of understanding of nature, many unexpected events such as natural disasters that were considered helpless in the past have become increasingly predictable. Only with relatively reliable predictions can better prevention and emergency management strategies be formulated.
Emergency management should be a "prevention first, prevention combined" strategy based on prediction and monitoring, rather than the hasty activation of the so-called emergency mechanism in many places when things go wrong. I have conducted at least 15 years of research and reflection on this aspect, and have written several articles, such as "Management of Festivals: Emergency Management is the Last Barrier to Safety Production" and "The Supreme Realm of Emergency Management is Without" Emergency "(available on Baidu)
A big challenge left for people now is earthquake prediction. I have written an article before titled "[Earthquake Prediction] Technically speaking, imminent earthquake prediction should be achievable" (96 readings, available on Baidu). In terms of the time span of earthquake prediction, there are several methods: long-term prediction, medium and short term prediction, and imminent earthquake prediction. Personally, I believe that imminent earthquake prediction is the most likely to achieve breakthroughs in technology, The problem lies in the government's review and release mechanism, mobilization ability, and the cultivation of public coping psychology.
I summarized the three major struggles of human beings over the past three million years. The Green Line is the integration of human beings and nature. From the Noah's Ark in the Western myth to the Dayu's flood control in the Chinese myth, the Dujiangyan Irrigation Project Water Diversion and Control Project built by Li Bing and his son in 256 BC, to the Three Gorges and the Yellow River Xiaolangdi Super Large Water Control Project, all of them reflect the history of human beings' struggle against nature. With the development of satellite remote sensing and telemetry technology, astronomical observation, and geophysical technology in recent years, humans have become increasingly capable of predicting and preventing natural disasters.
3.2 Apart from helpless natural disasters, the most hated thing is the sudden events caused by human disasters
In man-made emergencies, the vast majority are safety production responsibility accidents, and the vast majority of these responsibility accidents are related to equipment, for example, as I mentioned in the following article.
Lessons and Responsibilities from OTIS Elevator Accidents, July 2011
2. "Sanmenxia 719 Major Explosion Accident: Safety Awards Can't Cover up Sin", July 2019
3. Lessons from Traffic Accidents in Changhong Bridge, Beijing and Self reflection, June 2012
4. "A Sales Analysis of Lanzhou Petrochemical Explosion Accident", January 2010
5. "Reflection on Major Accidents in Shenghua Chemical Industry: The Revenge of People and Ghosts Responsible for Safety Production", December 2018
To eliminate safety production responsibility accidents, it is important to establish a safety production responsibility system and system with clear rewards and punishments, effective assessment, and strong supervision. At the same time, it is necessary to do a good job in the safe operation and maintenance of equipment, especially important equipment and facilities. If necessary, it is necessary to accelerate the construction of equipment operation and maintenance monitoring and monitoring systems based on Internet of Things technology.
I have too much experience in this aspect. Hold back and stop writing.
May Day holiday arrangements
Photos of the 2018 year-end group visit
NTT, QKL Ford to negotiate with our company
Picture of the meeting on March 9, 2019