In a unique way, quality of a finished product depends on an algorithm.
Designing, testing and optimizing prototypes for household refrigeration appliances requires an algorithm that controls all of the appliance’s components. Adjusting and modifying the control algorithm must be fast and effective, and the appliance’s various thermal and power responses must be measured and stored for analysis.
One quality solution: Using software and hardware to build a cost-effective automated control and measurement system that is easily adjustable and customizable for specific requirements. This solution makes developing new refrigeration appliances faster because we can control all of the appliance’s components and measure temperatures and energy consumption.
With the development of modern household refrigeration appliances, various components have been implemented to enhance functionalities such as no-frost, biofresh and multiflow. These components must be controlled in a way that optimizes appliance operation. A cooling appliance operates best when it achieves the desired temperatures at minimum energy consumption, without the loss of functionalities. Refrigeration appliances can be optimized in various ways including with long-term development measurements, complex mathematical models or an algorithm solution presented here. Long-term measurements are basically a trial-and-error approach, which is costly and time-consuming. Constructing mathematical models requires specific knowledge of expensive tools such as ANSYS Fluent or MathWorks’ MATLAB software, and is not necessary flexible, physical prototype changes that require construction of new models.
In our approach, National Instruments’ (NI) equipment measures specific types of thermal and power responses that are required for the simulation and optimization tool, an approach we developed in cooperation with the Jožef Stefan Institute in Ljubljana. Additionally, NI software continuously processes the acquired data for controlling the refrigeration appliance and stores it for later analysis. We then validate the obtained optimization results on the prototype using the same NI equipment and software.
Measuring the energy consumption of a refrigeration appliance requires at least 24 hours of stable operation, mainly because thermal processes are slow.
Taking Many Measurements
Each alteration of the appliance requires a new energy consumption measurement, which can take a lot of time. With the simulation tool, several days of refrigerator operations are simulated in a matter of seconds. The simulator results are based on a set of short response measurements, which have to be measured at a specific time and operation point. The automatic control and measurement system is based on data acquisition (DAQ) boards and the LabVIEW graphical programming environment.
It is the most suitable for performing such short measurements because it can easily adapt to special requirements, such as changing control modes, measuring responses at a specified time, or controlling the rotation speed of a compressor or fans.
The accuracy of the simulator depends on the quality of the measured short responses for different component state combinations. All the possible operating states must be measured in the same conditions and at the same operating point. Without concurrent control and processing of the responses with the application, defining the proper operation point would be almost impossible. When these measurements are complete and the simulator is operational, we can begin optimizing the appliance operations with genetic algorithms. The resulting optimized control parameters are then implemented on the actual prototype using the same equipment and software. The automatic control and measurement system is controlled with a LabVIEW application in Figure 1.
The application must control the temperatures of two separate compartments in the appliance: fresh food and freezer. Before starting the short response measurements, the appliance must reach the specified operation point. This can be difficult to achieve because the temperatures in the two separate compartments are not independent of each other and are also influenced by other factors such as time since the last defrost cycle, time since the last compressor cycle, and inner conditions in the cooling system. With automated response measurements, the time between two consecutive response measurements is shortened and diminishes the effects of the previous measurement on the cooling system. The measurement time was minimized by enough time that automatically there is a measure of all the possible states. With the graphical user interface capabilities of LabVIEW, inefficient states were examined and eliminated.
Temperature and Power Curves
When the short response measurements for all the states are complete (Figure 2), the temperature and power curves (data from the measurement system) are fitted with polynomial equations.
These equations are the basis on which the simulator then simulates the refrigeration appliance operations and calculates energy consumption. We estimate that the solution space for the refrigerator simulation contains up to 1,020 entities. Brute force searching for the best simulation would be irrational, but with genetic algorithms we can drastically reduce the search.
Optimization results, a set of predefined control algorithm parameters, are obtained within 24 hours on an octal core processor and then implemented and tested.
With National Instruments’ DAQ boards and the LabVIEW programming environment, the authors’ firm greatly reduced the time needed to optimize new prototype refrigeration appliances. The development measurements cost approximately $35 US or $26 per day per appliance, so the investment in such products was quickly returned.
The greatest improvement in energy efficiency on an existing appliance reached seven percent with this system. Appliances with improved energy efficiency and added functionalities give companies a competitive advantage.
A future goal is to use such solutions for standardizing and integrating control, measurements, simulation and optimization into one application to create a fully automated self-learning system for online optimization of control algorithms. Using machine-learning techniques, the most energy-efficient control algorithm can be found in the shortest possible time, with minimum user interaction.
Numerous Quality Outcomes
From washing machines to robots to home monitoring systems, quality and standards are making a design difference.
For staffers at Whirlpool Fabric Care, Advancement Development, the challenge was increasing reliability standards and testing capabilities of their washing machine electronic control boards and implementing an automatic system for embedded firmware validation to save time and resources.
Using National Instruments’ VeriStand real-time testing software and its PXI hardware, the solution creates a stand-alone system that tests and validates electronic control boards and automatically tests, calibrates and validates smart algorithms. Developed was smart control algorithms using rapid control prototyping systems based on the NI LabVIEW and its simulation interface toolkit. With company-wide use of this system, new concepts can be tested and developed with simulation tools directly on a real washing machine.
Developed in the washing machine mathematical model are main control loops inside the home appliance, such as mechanical, hydraulic and thermodynamic subsystems for simulation.
Then there are those robots.
The challenge centers on developing a prototype of an autonomous mobile robot for home monitoring and service that is equipped with two robot arms with multiple degrees of freedom to cater mainly to the elderly. And one solution is programming NI CompactRIO hardware to control several different devices at the same time, including two robot arms using a motion control algorithm; a light detection and ranging sensor for data collecting, navigation and obstacle avoidance; a mobile position-control algorithm based on encoder and data.
The inspiration is the home robot, designed to monitor home appliances through a network. It is especially useful for the elderly because its dual robot arms can fetch and deliver some items and can act as an intelligent wheelchair taking a person from place to place. The robot also possesses human-computer interaction functions such as voice and face recognition.
With the home monitoring system, “We used an NI USB-6008 DAQ device to acquire the signals from sensors. The software program developed in LabVIEW acquires the signals from the light detection sensor and the temperature sensor. After the signals are acquired, the software determines whether to turn on the light and fan. The light intensity and fan speed change depending on the light condition and temperature in the house. We used external hardware circuitry to control the light intensity and fan speed through the DAQ device,” says Daniel Liew Wei Chao of the Temasek Polytechnic.