Anyone who has studied the field of science and engineering should have heard the word “control” once. In particular, “automatic control” is often used in the field of science and engineering. The field of control is so deep that one specialized book can be written by itself, but this time, let’s take a brief look at “What is control?”. We will discuss the difference between automatic control and manual control, the difference between feedback control and feed-forward control, and the relationship between artificial intelligence and edge computing, which have become popular in recent years.
What is control? Automatic control and manual control
What is the definition of control? “Control” is defined as “manipulating and adjusting the system to achieve the desired state”. That is, to work with the system to bring it to the desired state or to maintain it.
“Control” can be roughly divided into two types. “Manual control” and “automatic control”. Manual control means that humans work on the system. In other words, it is defined as “human intervention in the operations and adjustments performed on the system.” For example, if it’s cold and usually a bonfire, and the fire is so small that it doesn’t get warm, then “control” is to increase the amount of firewood to make the fire bigger. In such a case, it is considered that human beings control the size of the bonfire by adjusting the control amount of “the amount of firewood” with the “size of fire” as the target amount for the “heating system” of the bonfire can do.
On the other hand, automatic control is automatically working on the system. In other words, it is defined as “performing operations and adjustments to the system without human intervention”. For example, when it is hot and the air conditioner is turned on, humans only set the temperature. Then, the air conditioner automatically adjusts the temperature to reach or maintain that temperature. In such a case, it can be considered that the “cooling system” called an air conditioner automatically controls the temperature with the set temperature as the target amount and the refrigerant circulation amount as the control amount.
Classic control method, feedback control, and feedforward control
In the above air conditioner example, the amount of refrigerant circulation is controlled as the control amount for the set target amount. In other words, the current amount (room temperature) is measured by a sensor, and the controlled amount (refrigerant circulation amount) is determined by comparing it with the target amount (set room temperature). And by repeating this, the current amount is gradually approaching the target amount. In other words, the control method is used to bring the current quantity closer to the target quantity by comparing the current quantity with the target quantity and adding the difference to the current quantity. Such a control method is called feedback control. This is because the difference is fed back to the current amount and added.
Feedback control is a very common and widely used control method. However, in principle, it has the disadvantage that there is a delay before the current amount reaches the target amount. For example, it takes time to reach the set temperature of the bath, and it takes time for the temperature of the air conditioner to stabilize.
On the other hand, there is feedforward control. Feedforward control is a control method that predicts the occurrence of a disturbance and the amount of control when there is a disturbance in the system or when the required amount of control can be predicted in advance and adds the corresponding amount of control. Often used in addition to feedback control. For example, feedforward control may be used to determine a rough control amount, and feedback control may be used for fine adjustment. Artificial intelligence may also be used to predict the amount of control.
As a specific example, instead of measuring the current temperature of the bath and comparing it with the controlled amount, it is conceivable to predict it in advance and output the controlled amount commensurate with it. This shortens the time it takes to reach the set temperature.
Automatic control and artificial intelligence
Some recent air conditioners and baths automatically set the optimum temperature when you press the “Random button”. Such control cannot be achieved by classical feedback control or feedforward control alone. This is because it is necessary to collect multiple factors at the same time to determine the control amount in order to determine the optimum temperature. For example, when setting the temperature of a bath, it is necessary to collect factors such as weather, temperature, and humidity as well as the temperature of the bath to make a comprehensive judgment and determine the amount of control of the bath.
In addition, when using household fuel cells, which have become widespread recently, it is necessary to determine the optimum control amount by considering the amount of power generation and the remaining amount of the tank in addition to the hot water temperature. Therefore, there are more factors to consider than traditional water heaters. And the factors that need to be considered tend to increase as new technologies become available. Under such circumstances, it is becoming necessary to perform not only control over the target value but also “optimal” control over the entire system.
As an answer to such “optimal” control requirements, artificial intelligence, specifically deep learning by neural networks, is often introduced to make comprehensive decisions. Neural networks can determine the amount of control by comprehensively judging multiple factors at the same time. In addition, the tremendous development of hardware and software in recent years has opened up bright prospects in terms of speed. Although learning is required, the benefits of being able to determine the optimal amount of control for a large number of factors are enormous. As a technology that complements conventional feedback control and feedforward control, it will become indispensable for future control technology.
Automatic control and edge computing
In the above example, we gave examples of air conditioners and water heaters that do not require much control speed, but even in these examples, it is better to have a high control speed. So what about other areas? Especially in the fields of industrial systems and automobiles, more speed is required for control. Taking the engine control of a car as an example, the control is performed in units of a few milliseconds. Similarly, in production systems, control systems require speed. Since the advent of cloud computing, the use of the cloud has been increasing in various industrial fields, but the bottleneck is its speed. It is not easy in terms of speed to realize the control system mentioned above only by cloud computing. Therefore, edge computing, which is advantageous in terms of speed over cloud computing, is considered to be effective.
The future of automatic control
Automated control is actually a very long-standing field that is said to have been mechanically realized by the end of the 18th century. In the old days, these automatic controls were realized in analog form. The classic control method was established in this analog era. Currently, this is digitized and used, but from now on, artificial intelligence will complement it. Automated control, which has gradually changed its shape by incorporating new technologies, is still being updated by technologies such as artificial intelligence and edge computing.