Applications of machine learning and other forms of artificial intelligence have been recognized in robotics and analytics. Now the technology is adding some spice to basic control applications.

Using your noodle to think things through tends to make things go much more smoothly—even if you’re just a high-speed food packaging machine wrapping instant noodles. That’s an important lesson gained from machine learning technology used by systems integrator Tianjin FengYuLingKong of Tianjin, China.

This form of artificial intelligence (AI) allowed the firm’s engineers to develop a multivariable inspection model for one of China’s largest producers of noodles. Relying on this model, the control system for the packaging lines can now deduce whether sachets containing spices and dried vegetables for flavoring were placed correctly on the precooked noodle blocks before each block is individually wrapped.

This ability is an example of how machine learning and other forms of AI are moving beyond applications like robotics and analytics and into control applications.

In Tianjin FengYu’s case, there was no other cost-effective way to check whether an occasional sachet of flavorings might have slipped between two blocks of noodles and been cut open by a cross-cutting tool. Although cutting a sachet generates measurable signals within the machine, other events such as vibration and changes in packaging material, conveyor speed, and cutting tension also affect those signals, making conventional forms of process monitoring unreliable.

For this reason, Tianjin FengYu decided to develop, train, and deploy a mathematical model using TwinCAT Machine Learning from Beckhoff Automation. The integrator’s engineers collected sensor data via EtherCAT terminals and TwinCAT Scope View charting software. Then, the data were correlated into a model using TwinCAT Condition Monitoring, and the model was trained using an open-source framework called Scikit-learn.

After being saved as a description file in a binary format suited for serialization in TwinCAT, the trained model was loaded into a CX5100 series embedded PC, which runs the model in real time. This embedded PC is integrated with the main controller on the packaging line.

The control system can run the model in real time as each packaging line wraps about 500 packages of noodles per minute. “A trained model actually runs fairly quickly,” notes Beckhoff’s Daymon Thompson. “And that’s what’s usually running in the controllers.”

Training the model is a different story, however. Thompson says that training needs a lot of processing power, as much as 30 minutes to a full day, depending on the model and the computer training it. So, the initial training and any subsequent retraining are often done on a server or an offline controller.

Besides in-process inspection, another application for machine learning in controls is the optimization of motion profiles. Consider a conveyor system that carries payloads around corners and coordinates motion with loading and other activities in a demonstration created by Beckhoff using its eXtended Transport System (XTS). “Instead of just running everything around as fast as we can to get in line for the next synchronized event, we want the motion to be optimized to minimize energy consumption and wear and tear on the mechanics,” explains Thompson.

The machine learning algorithm figures out exactly what the motion profile should look like. “Because the motors driving the system need to be coordinated in real time, the motion profile really needs to be built into the machine control,” notes Thompson. “It can’t be done on a server or even an edge device.”

AI benefits closed-loop control 
“Traditionally, PLC programmers would write ladder logic to tune systems with either creative rungs of arithmetic or PID control blocks,” says Kevin McClusky, co-director of sales engineering at Inductive Automation. “Today, closed-loop control with AI allows users to feed data into predictive models that can optimize output based on past performance or cost reduction, allowing far more complex algorithms to be applied to achieve efficiency or productivity goals.”

He reports that the catalogs of several PLC manufacturers now offer AI modules for closed-loop control. Although not every application needs the technology, these modules are another set of tools in the toolbox. McClusky compares them to a simple PID block in ladder logic. “It’s not needed in a lot of applications, but it sure is handy in applications that can benefit from it,” he says.

“Model outputs can be integrated into the control scheme to extend the capabilities of classical control methods,” adds Jennifer Mansfield, marketing manager—analytics at Rockwell Automation. “Challenging problems, like enabling predictive maintenance or dynamic control, are better addressed with machine learning than classical control.”

Illustrating her point is the model predictive control (MPC) that EnWin Utilities Ltd. implemented to mitigate pressure spikes in the water distribution system in Windsor, Ontario. These spikes had been contributing to an increasing number of watermain breaks in the aging system.

The old control scheme had depended upon PID logic that maintained a flow setpoint based upon outlet header pressure. Pressure would vary whenever operators would start and stop pumps at the two pumping stations and an auxiliary booster station to adjust flows to compensate for fluctuating demand.

To even out pressure, EnWin chose an MPC-based system that could handle more variables than just flow and outlet header pressure. Working with engineers from Rockwell Automation, EnWin began by creating 17 remote pressure stations throughout the water distribution system. The team also installed server-based MPC on its existing supervisory control and data acquisition (SCADA) system. The system now maintains the lowest possible pressure for producing adequate flow as demand changes.

To optimize pressure and flow control further, the main campus uses a new ControlLogix controller with onboard MPC. “We knew we could optimize the system by incorporating pump start-stop functionality and flow control valves,” explains Quin Dennis, an application engineer at Rockwell Automation. “But given the existing interval speed, [server-based] MPC would not be able to make system adjustments quickly enough to mitigate the rapid pressure spikes from pump starts or stops.” Onboard MPC, however, reduced the 15 to 16 second interval speed down to 0.5 to 1 second.

The system is now responsive enough to regulate the speed of the pumps and adjust control valves to offset any pressure spikes. By replacing PID logic with MPC at the controller level, as well as at the server level, EnWin was able to reduce watermain breaks by 21%. It also reduced average pressure by 2.8 psi and standard deviation by 29%, saving the company $125,000 in annual energy and leakage costs.

Predictive applications
Another benefit of AI is that it can help users peer deeper into their processes than controllers would otherwise permit. This is especially true in applications that require processing large amounts of data.

“AI is now being implemented on the edge in situations where large volumes of data must be analyzed quickly before being sent to the cloud,” observes Joe Berti, vice president of AI applications atIBM Corp. “As a result, smart technology is broadening engineers and technicians’ understanding of their assets’ health by capturing and interpreting more information faster than any human could.”

Consequently, Berti thinks that the biggest contribution AI and machine learning are making to controls technology is the ability to streamline detection and resolution of developing problems before they have a chance to escalate. “In the past, an asset might have been inspected on an annual basis,” he says. “Now, IoT sensors and enterprise asset management systems can detect patterns in asset data and then translate those findings into potential problems.”

An example of this kind of application can be seen in the use of AI to discover oil degradation on a food packaging line installed by Novate Solutions Inc., an engineering and technology services firm based in West Sacramento, Calif. The clues to the problem came from IBM’s cloud-based AI technology and Maximo Monitor software, which Novate uses to provide a process monitoring service. The AI noticed that the average torque of a servomotor had been increasing over time, so the Novate solution flagged the equipment for inspection.

Upon being alerted by Novate engineers, the production crew at the food producer checked the packaging equipment and found that the oil had not been changed as the maintenance log had suggested. The oil had completely degraded, causing the motor to work increasingly harder over time.

Trained for decision-making
Another application for AI in basic control is the automation of decision-making in continuous processes. “Here, an AI system controls a part of a facility or operation, sending signals to do basic control of different pieces of equipment,” says Inductive Automation’s McClusky.

He points to the way a type of machine learning known as reinforcement learning is being deployed by Andritz Automation, a worldwide systems integrator headquartered in Graz, Austria. In reinforcement learning, models are trained to make a sequence of decisions by means of a trial-and-error method that strives to maximize a cumulative score of rewards and penalties.

In what may have been the first implementation of this AI technology in continuous industrial processes, Andritz engineers in Canada and Germany collaborated on developing prototype software. They then implemented the prototype in a pilot program at Newmont GoldCorp., a Vancouver-based goldmining company.

This prototype uses the integrator’s process simulation software as the training ground for machine-learning algorithms. The AI engine learns by interacting with several simulations as they run. A user can set up batches of training scenarios, such as particular plant malfunctions that the AI engine needs to know. After the training exercises are performed, the algorithms are stored and used for automatic plant control.

A key technology for developing and implementing this AI engine was Inductive Automation’s Ignition development environment for SCADA. Ignition provided a bridge between the AI engine and either the integrator’s process simulation software or the real plant, using scripted HTTP calls on one side and OPC on the other. 

Ignition’s sequential function charts control the dispatch of training scenarios. All scenario configuration and training results are stored in a SQL database. During the training process, the two teams in Canada and Germany were able to work on the project at the same time because the training environment was deployed on a small virtual network on a Microsoft Azure cloud server in Europe. Each team could run Vision clients simultaneously and access the database gateway and simulation machines.

The post When Artificial Intelligence Comes to Control appeared first on Artificial Intelligence.

This is also the case in the quality control department of a French manufacturer of plastic caps. Reliably detecting cracks and micro-cracks on plastic caps in 40 different colours and shades running at high speed on a production line is a real challenge.

APREX Solutions from Nancy, France has successfully achieved this goal with the help of image processing technology and artificial intelligence. The basic images are provided by a USB 3 industrial camera from IDS Imaging Development Systems.

SOLOCAP is a subsidiary of La Maison Mélan Moutet – “flavour conditioner since 1880” – and manufactures all types of plastic caps for the food sector at its industrial site in Contrexéville.

Among them, a top-class screw cap suitable for any glass or PET bottle. Thanks to a clampable lamella ring arranged around the bottle collar, it enables simple, fast, absolutely tight and secure seal.

However, the slats must be reliably and extremely carefully checked for cracks, tears and twists during production. This is the only way to guarantee absolute tightness.

The previous inspection system could not meet these high requirements. APREX Solutions realised the new solution with artificial intelligence individually on the basis of in-house software algorithms.

The necessary specifications were developed in advance in cooperation with the customer. This also included several inspection stages, one of which, for example, was the reject control to avoid false reports. The introduction took place in two phases:

First, the specific “SOLOCAP application” was trained with the help of the intelligent APREX Track AI solution. The software includes various object detector, classifier and standard methods that operate at different levels.

Networked accordingly, they ultimately deliver the desired result tailored to the customer. Four control levels with several test points guarantee a reliability rate of over 99.99 percent.

In the second step, this application was implemented in the production line right after the first assembly run with APREX Track C&M. The latter was specially developed for the diverse image processing requirements in the industrial sector.

This includes, among other things, the control and safeguarding of a production line up to the measurement, identification and classification of defects in the production environment.

The software suite delivers the desired results quickly and efficiently, without time-consuming development processes. After a short training of the AI methods, the complete system is ready for use at the customer.

In the case of SOLOCAP, it combines an IDS UI-3280CP-C-HQ industrial camera, powerful ring illumination and a programmable logic controller (PLC) to provide comprehensive control over all inspection processes.

At the same time, it records all workflows in real time and ensures complete traceability. Only one camera is needed for this. However, APREX TRACK C&M could handle up to 5 cameras.

Romain Baude, founder APREX Solutions, says: “The difficulty of this project consisted mainly in the very subtle expression of the defects we were looking for and in the multitude of colours. With our software suite, it was possible to quickly set up an image processing application. Despite the complexity.”

The image from the camera provides the basis for the evaluations. It captures every single cap directly in the production line at high speed and makes the smallest details visible to the software.

The UI-3280CP-C-HQ industrial camera integrated into the system with the 5 MP IMX264 CMOS sensor from Sony sets new standards in terms of light sensitivity, dynamic range and colour reproduction.

The USB 3 industrial camera provides excellent image quality with extraordinarily low-noise performance – at frame rates up to 36 fps. CP stands for “Compact Power”.

This is because the tiny powerhouse for industrial applications of all kinds is fast, reliable and enables a high data rate of 420 MByte/s with low CPU load.

Users can choose from a large number of modern CMOS sensors from manufacturers such as Sony, CMOSIS, e2v and ON Semiconductor with a wide range of resolutions.

Its innovative, patented housing design with dimensions of only 29 x 29 x 29 millimetres makes it suitable for tasks in the fields of automation, automotive, medical technology and life sciences, agriculture, logistics as well as traffic and transport, among others.

Screwable cables ensure a reliable electrical connection. Thanks to the IDS-characteristic plug & play principle, the cameras are automatically recognised by the system and are immediately ready for use.

Romain Baude says: “The excellent colour reproduction of the UI-3280CP-C-HQ and its high resolution of 5 MP were decisive factors for us in choosing the camera. At the same time, the model enabled a quick, uncomplicated integration into our system.”

Outlook

Anthony Vastel, head of technology and industry at SOLOCAP, sees a lot of potential in the new inspection system.

He says: “APREX’s AI-based approach has opened new doors for our 100 percent vision-based quality control. Our requirements for product safety, but also for reject control, especially in the case of false reports, were quickly met.

“We are convinced that we can go one step further by continuing to increase the efficiency of the system at SOLOCAP and transferring it to other production lines.”

AI offers quality assurance, but also all other industries in which image processing technology is used, new, undreamed-of fields of application.

It makes it possible to solve tasks in which classic, rule-based image processing reaches its limits. Thus, high-quality results can be achieved with comparatively little effort – quickly, creatively and efficiently.

APREX Solutions and IDS have recognised this and offer solutions with intelligent products that make it easier for customers to enter this new world. Image processing and AI – a real dream team on course for growth.

The post Camera-based cap control with artificial intelligence appeared first on Artificial Intelligence.

Fraunhofer Institute scientists have developed quality control software that allows users without expert AI knowledge to benefit from AI.

Although they can bring great benefits in everyday work, many small and medium-sized enterprises (SMEs) shy away from applications based on artificial intelligence. But AI offers a lot of potential, especially in quality control. Nevertheless, training the models is difficult and hardly feasible without mathematical knowledge, as there are countless parameters that can go into such an analysis. And once an AI algorithm is learned, it is trained only on the specifications it learns. If a product design or the geometry of a component is later changed even slightly, the algorithm recognizes this as an error and the AI must be retrained.

Researchers at the Fraunhofer Institute for Digital Media Technology IDMT in Ilmenau, Germany, have now developed the “IDMT-ISAAC” software, which can be operated even without extensive expert AI knowledge. IDMT-ISAAC stands for Industrial Sound Analysis for Automated Quality Control. “We want to enable SMEs to adapt and customize AI algorithms themselves,” says Judith Liebetrau, group leader of Industrial Media Applications at Fraunhofer IDMT. “They can apply IDMT-ISAAC to their own audio data, retrain it, and thus get fast and reliable results and decision support for their quality assurance.”

Using AI even without expert knowledge

IDMT-ISAAC relies on acoustics for analysis, since in many cases it is possible to detect defects just by the sound of the process. To train the AI, the scientists use recorded acoustic data from welding processes. The AI analyzes the typical noises that occur and draws conclusions about the quality of the respective weld seam from the audio data. If, for example, the geometry of a product is then changed, the user can teach this to IDMT-ISAAC with just a few clicks. As early as summer 2021, the software should be adapted to live operation to the extent that the system can immediately analyze real-time data from production and optimize quality assurance. In three to four years, it should even be able to actively intervene in production.

But the framework at the heart of IDMT-ISAAC doesn’t offer new analysis options just for welding. “We have integrated various methods in the modular system to be able to map other processes, such as milling, relatively quickly,” Liebetrau explains. Companies that already have their own software should also be able to use it in the future. They will also be able to access the institute’s AI via an interface on the Fraunhofer IDMT server. It is important to the developers here to emphasize that data protection and data security would always be observed and that the data would be processed anonymously, regardless of whether companies access the AI via an interface or it is integrated into the company via the framework.

Making AI comprehensible

For different user groups – AI novices as well as AI experts – the software can be customized via different user profiles. For example, developers of AI algorithms are very interested in getting a feel for how AI makes its decisions and the sounds it uses to make them, says Judith Liebetrau. “So we are also moving a bit in the direction of Explainable AI with the framework to make AI more comprehensible,” Liebetrau says.

The researchers will present IDMT-ISAAC at the Hannover Messe from April 12 to 16, 2021. At the virtual booth, Bescher will apply artificial intelligence models using the IDMT-ISAAC software to industrial audio data to verify its quality.

The post Acoustic Quality Control with the Help of Artificial Intelligence appeared first on Artificial Intelligence.

As climate change heightens the threat of flooding, a big data and cloud computing company in Guizhou province is intent on harnessing a range of information to track risk factors and send alerts in a rapid and cost-effective manner.

The conventional method for predicting floods relies on setting up physical hydrometric stations in high-risk areas. Based on information transmitted from sensors placed near waterways, a number of variables are monitored, including water level, flow rate, temperature, rainfall and evaporation, and then analysis is conducted accordingly.

The method is effective and essential, but the caveat is that to increase its precision inevitably requires investing more in infrastructure, said Li Tao, chief engineer at Guizhou East Century Science& Technology, a company founded in 2000 in Guiyang, capital of Guizhou province.

With extreme weather shifts becoming more common, it is not feasible to continuously establish new hydrometric stations in every area deemed as flood-prone, Li said.

The company, set up in the city regarded as the epicenter of China’s emerging big data scene, has instead looked to the digital sphere to tackle mounting flood hazards.

The novel warning system, known as East Auspicious Clouds, first aggregates meteorological and geographic information, as well as data on water bodies and rainfall, from different government departments and research institutions.

Both real-time and historical data are fed into an analysis model that boasts one of the fastest speeds in the world to complete one session of comprehensive analysis of all data at hand.

“The system breaks from the traditional flood forecasting mechanism that usually costs a lot in terms of construction and equipment maintenance. It also covers a much wider range of areas and achieves a high precision rate,” Li said.

“The traditional sensors measure the amount of water on the ground. By comparison, the system begins calculating and forecasting the amount of rainfall when raindrops begin forming from water vapor in the air,” he said.

Yu Linmei, deputy general manager of the company, said in an earlier interview that the system is able to renew its alerts every 15 minutes. The normal speed for domestic counterparts is one hour to 90 minutes, and for most advanced global competitors about 30 minutes.

On average, the system is able to issue warnings for small and medium-sized bodies of water about one to four hours in advance. The window of opportunity for early preparedness is estimated to reduce economic losses by up to 90 percent, according to the company.

The idea of tapping into big data’s potential for flood prediction work was first hatched in 2014. The first version of the system was rolled out in 2018 and the second in May 2019.

So far, the system has been mainly applied in Guizhou province and is also making headway in other parts of the country, the company said.

In June, the system had successfully predicted 10 hours in advance that a stretch of railway tracks connecting Guizhou and Sichuan provinces were at risk of being submerged.

The information enabled the railroad authority in Chengdu, capital city of Sichuan, to take precautions and minimize the flood’s disruption to regular rail service.

“Imagine how many more physical sensors are needed to obtain such information. For the system to sharpen its sensitivity and accuracy, the key is to gather as much data as possible. The more the data, the better the outcome,” Li said.

However, he added that the system at the moment only acts as an additional source of information for decision-making provided to those responsible for gathering insights from various parties before taking adequate precautions.

Predicting natural disasters is extremely complicated. None of the experts in the industry can assert that their projections are absolutely correct, but it is always helpful to see the issue from another perspective, he said. Li added that it will take time for flood-prevention workers and authorities to learn and adopt a brand-new system that overturns the conventional method and uses novel technologies such as big data and cloud computing.

The post East Century uses big data for flood control appeared first on Artificial Intelligence.

Rethink, now part of the Hahn Group, says the ASM setup was simple and the operation enabled Sawyer to “quickly achieve high levels of acceptance among employees”.

The task of the SMT Solutions business segment in the ASM Pacific Technology group is to support, implement and realize the SMT Smart Factory for electronics manufacturers worldwide.

ASM solutions such as the SIPLACE Placement Systems and the DEK Printing Solutions support the networking, optimization and automation of central workflows at the line and factory level with hardware, software and services and allow electronics manufacturers the gradual transition to the Smart SMT Factory with dramatic improvements in key figures / KPIs for productivity, flexibility and quality.

The central strategy element at ASM is the close cooperation with customers and partners. ASM initiated the SMT Smart Network as a global competence network for the active exchange of experiences between Smart Champions.

ASM is a co-founder of the ADAMOS joint venture to develop an IIoT platform for manufacturing companies and, together with other SMT manufacturers, is establishing the open standard HERMES as SMEMA successor for M2M communication in SMT lines.

The task

ASM uses Rethink Robotics’ cobot Sawyer in quality control. Sawyer independently performs the checks by picking up a part from the existing assembly, performing a scan of the surface and placing the board in a test adapter.

After this OK / NOK test, Sawyer separates the parts based on the results and releases them for further processing. Defective parts are removed from the process chain and deposited by Sawyer in a special area so that they can be transferred to the fault analysis.

Another challenge initially was the employees’ acceptance of the cobot. Manufacturing workers feared that Sawyer would be a machine that could endanger their jobs in the long term.

However, this idea changed quickly because Sawyer takes on repetitive tasks, which means that ASM employees can apply their skills for more demanding tasks in the value chain.

After a brief familiarization, the cobot has become part of the team in production, which gave him the new nickname “Paul”.

Project progress

The ASM decision-makers got to know a Sawyer application, which offered an opportunity to automate the previously manually performed quality control of printed circuit boards. Sawyer was able to pick up parts independently and use them in a test system.

During the first meeting for the live demo of Rethink Robotics, the ASM decision-makers were impressed by how easy it is to program Sawyer through the Intera software.

Advantages for the customer

Alessandro Bonara, head of electronic manufacturing at ASM Assembly Systems, says: “It was very impressive for us that the Sawyer is so quick and easy to assemble.”

After delivery, assembly and connection, Sawyer is immediately ready for use. In comparison to conventional industrial robots, Sawyer is much easier to set up and put into operation.

Another bonus is that Sawyer does not have to be fixed to the floor of the assembly hall with screws, which makes its mobility another huge advantage in terms of flexibility.

Programming the Intera software is very easy for the user due to its structure and operation.

Employees can either build the program modularly using “Teach by Demonstration”, which works by moving the robot arm in the room, or using landmarks to specify a framework that Sawyer can use to learn how to perform his movements.

As a final step, an employee can fine-tune the programmed movements using the Intera software.

The safety aspect in the cooperation between humans and collaborative robots has the highest priority when using Sawyer.

That is why cobot Sawyer continuously controls his use of force, the position of his robot arm and has a collision control. The entire application used in production at ASM is CE certified.

Productivity and quality improvements

The cooperation between Rethink Robotics and ASM was “very constructive and reliable”, say the companies.

For ASM, productivity and quality improvements are among the most important goals.

The use of technologies for automation is an important step, which is why the automated quality control with cobot Sawyer fits well into the ASM manufacturing process.

The post ASM Assembly Systems uses Rethink Robotics’ Sawyer robot for quality control checks appeared first on Artificial Intelligence.