Data Science and PID: the key to an efficient industry
Copyright Crank AMETEK
The Proporcional-Integral-Derivative (PID) was created in 1911 by Elmer Sperry, initially for the use of the United States Navy. The goal was to automate the direction of ships emulating the behavior of a helmsman capable of compensating for constant variations on the seas. After a few years, engineers published the first theoretical analysis of this control, describing it through a mathematical equation that is a fundamental concept in control theory.
PID is one of the most widely used control techniques in industry, with the role to ensure the stability and performance of production processes. Whether in temperature, pressure, flow or level regulation, PID stands out for its versatility and efficiency. Each component compose a precise control of the system, they are:
- Proportional Control (P): The proportional control generates an output signal that is proportional to the current error, which is the difference between the required value (setpoint) and the measured value (process). This action helps reduce the error occurrences, but only itself does not eliminate it.
- Integral Control (I): The integral control produces an output signal that is proportional to the accumulation and duration of the error, thus eliminating the persistent error.
- Derivative Control (D): The derivative control acts on the rate of error change, which helps to predict its future behavior and make quick adjustments. This way, it is possible to improve the stability of the system and reduce its oscillation.
However, the use of these controllers can still be improved, mainly by combining them with innovative technologies such as data science. The lack of visibility into the performance of PIDs and the difficulty in identifying opportunities for improvement can lead to issues such as process instability, wasted energy, and high maintenance costs.
Visibility that creates more assertiveness in industry
The PID controller can be graphically represented, enabling the visualization of system functioning over time. The graphic contains:
- X-axis: The horizontal axis usually represents time;
- Y-axis: The vertical axis represents the value of the controlled variable, such as temperature, pressure, velocity, etc.
- Setpoint Line: A straight horizontal line that represents the required value (setpoint) of the controlled variable.
- Controller Output: A curved line that shows how the controlled variable changes over time in response to the PID control. It starts at the initial value of the controlled variable and moves until it eventually stabilizes at the setpoint value.
In addition, it is important to note that if the controller output exceeds the setpoint value, the process is called an “overshoot”. If it falls below the setpoint value, it’s called an undershoot. The PID controller is rightly used to minimize these variations.
Finally, the performance evaluation of PID is done considering some different aspects. Rise time is the name of the response time of the system when it reaches the setpoint value for the first time. Settling time is the time it takes for the system response to stabilize at the setpoint value. The steady-state error is the difference between the setpoint value and the final value of the system response.
The figure below is a graphical example of PID controller action:
Copyright ST-One
The PID Controller in industry
The PID controller is an essential tool in industrial automation. According to the Federal Institute of São Paulo, in the article “PID Controller and Its Applications” (2021), about 97% of industrial applications make use of it. It is usually connected to the Logic-Programmable Controllers (PLCs) and is designed for specific control tasks that require often and continuous modulation.
In industry, PID can be used to control different kinds of variables, such as:
- Temperature control in industrial furnaces and machines, through the work of temperature sensors;
- Monitoring the liquid level in storage tanks, adjusting the in and out flow rates;
- Pressure control in a hydraulic system, by setting the pump or relief valve within the desired range;
- Adjustment of machine speed, configuring the delivery of supplies within the established parameter;
However, extracting information from a PID controller in a PLC presents some challenges. To make this process happen, it is necessary to establish balance between the parameters, that is, to find the right values for the proportional, integral and derivative constants. This requires in-depth technical knowledge of the systems and careful calibration. Also, it is necessary to know how to clearly identify which variables are controlled and which are manipulated and identify errors when there is a difference between them and the required value.
Therefore, to make full use of all functionalities of the PID controller, the industry must collect and analyze the machinery data. The data science solutions gather data from sensors and connected devices to monitor system performance, identify gaps, and optimize the production process.
Copyright ST-One
PID and Data Science results in more optimization
It is possible to use statistical techniques to collect and analyze data from PID controllers in real time. This analysis enables to identify patterns, trends, and anomalies that can affect control performance. Based on this information, it is possible to adjust the PID parameters precisely and efficiently, ensuring a more stable, accurate, and low-cost control.
Among the metrics used to evaluate the performance of PIDs, there are:
- Maintenance Score: Indicates the need for preventive or corrective maintenance of the controller based on indicators of presence of control saturation.
- Fit Score: Evaluates the quality of the PID fit, considering stability, speed of response, and steady-state error.
To the analysis be intuitive, the graphics that represent its calculation need to be clear and customizable, making it easy to see and understand the information. With adjustable elements, the dashboards anable operators and engineers to monitor PID performance in real time, identify problems, and make quick and assertive decisions.
In this context, with data science it is possible to obtain:
- PIDs Performance Overview: Allows you to track the performance of all industrial plant controllers in only one spot;
- Quick problem identification: Instant alerts and notifications about setpoint deviations, control instability or maintenance needs;
- Comparative analysis: Makes it possible to compare the performance of different types of controllers, from best to worst, and implement improvements;
- Data track history: Allows you to monitor the evolution of PIDs performance over time and evaluate the impact of optimization actions;
More accurate performance, by collect data in real-time
An industry implemented PID controller on a daily basis differs positively from the one that doesn’t. By its use, it is possible, for example, to have a better error control between the desired value and the monitored value of a system variable. Also, the production line that counts on this type of controller answers more effectively to changes in production environment and adjusts quickly to keep the controlled variable close to the desired value.
As a consequence, the whole industry suffers an improvement in Operational Efficiency, especially with the collection and analysis of data from a system with a PID controller. This great scenario is a reality because it is possible to evaluate all the Parameters of the controller (proportional, integral, and derivative, as seen) and its performance over time. The access to this information provides an overview of the operation, which serves as a basis for more assertive improvements.
The use of data science represents a true revolution in PID control. Through the union of these two technologies, it is possible to have more balance between the parameters, a flexible control, and simulations of the system future behavior. By providing accurate, real-time information on PID controller performance, data enables industries to optimize their processes, reduce costs, and increase efficiency.
Learn more about us.
