LEC MCheck

LEC MCheck is a completely new software solution for the diagnosis of sensor errors for test beds with frequently changing test carriers, such as engine test beds in research and development. LEC MCheck detects faulty measurements at an early stage and is able to identify faulty sensors or measuring devices, which not only improves the quality of the measurement data but also increases the efficiency of the test bed.

LEC MCheck Key Features

Automated system for measurement data plausibility check and error detection of measured values
Identification of the faulty sensor
Verification of the measurement data quality by means of different physical and statistical methods
Automatic generation of data-based models
Dynamic determination of confidence intervals and limit values for plausibility criteria
Determination of the monitoring quality
Software updates and further development

Get in touch

Methodology
Application example
Dashboard

In the first step, symptom generation, the test bench is mapped using physical models available for specific components in a model library and data-driven methods. Each model provides a residual based on the deviation between measurement and model-based calculation, which serves as an error indicator.

Based on the residuals, the second step of the diagnosis determines whether an error occurred during the measurement by checking the error conditions.

In the last step, error probabilities are calculated for all channels and sensors considered using a geometric classification procedure to determine which sensor is faulty in the event of an error.

Methodology

LEC-MCheck is a diagnostic system that can be used to check measurement data obtained on test benches for plausibility. The underlying methodology essentially consists of three steps: symptom generation, fault detection and fault isolation.

Model Library
selected modules are shown in the test bed scheme

User Defined Models
are used additionally, basically for simple plausibility rules

Data Driven Models
are used additionally, model training is done online and automatically

Diagnosis Results
detection and isolation rate for different fault intensities over all different sensors shown in the test bed scheme

Application example

Due to the modular structure, it is possible to adapt LEC-MCheck to different test bed systems. The components of the model library can be combined in any way to represent the test bench system. In this example, for example, a single-cylinder test stand is equipped with different components. For example, cylinders, piping, throttles, and even the entire system can be represented with specific models.

Data-driven models are automatically formed on the basis of historical data. Simpler checks such as limit checks of individual measuring channels can also be integrated. Further expert knowledge can be integrated into the system using user-defined formulas. This fully flexible design is one of the software’s key features.

Simulated faults can be used to analyze the monitoring quality of the system. For this purpose, the average percentage of correctly classified and correctly isolated faults is calculated for different fault intensities. The more intensive the fault is, the sooner it can be detected by the system and assigned to the correct sensor.

Fault Probability
MCheck identifies faulty sensors

monitoring quality
the higher, the better a channel can be monitored by MCheck

model based diagnosis
list of models for every used tool

basic tools
can be used independently of the type of test bed

system specific tools
– comprehensive model library for engine test beds
– expandable for non engine applications

traffic light
simple visualization of measurement data quality

Graphical user interface

The user interface of LEC-MCheck is divided into three levels. On the left is a list of all monitored channels and sensors. This can be both time-based and crank angle-based measurement data. On the right are the models used for fault diagnosis. These include the basic models such as limit checks or data-driven models, but also the test bench-specific models of the individual components. In the center of the interface, the currently selected elements are displayed, which can also be configured here.

LEC-MCheck has a configuration mode on the one hand and a diagnostic mode on the other. In the configuration mode all settings for the diagnosis can be managed. If the configuration is valid, it is possible to switch to the diagnostic mode. In this mode, the measured points are evaluated automatically. The result of the error diagnosis is initially displayed clearly by means of a traffic light. More detailed results, such as monitoring quality in the fault-free case and fault probability in the faulty case, or even the residuals of the individual models, are displayed in the user interface if required.

LEC MCheck in practice

Combustion engines have now reached a very high level of development, so that the potential for further improvements is becoming ever smaller. In the field of large engines, for example, individual technological measures can often only achieve improvements in the efficiency range of a few tenths of a percentage point. In order to enable the reliable evaluation of such measures during development in the first place, extremely high demands are placed on the measurement technology used. In general, it can be said that in engine development, a significant cost factor arises from experimental investigations on the engine test bed.

Late detected faults cost 40 percent test bed time

In order to keep the test duration and the resulting costs low, it is necessary to maximize the significance of test results by ensuring the highest possible measurement data quality. Faulty measurements are often only discovered during post-processing and lead to results that are difficult to interpret, and thus often to the need for expensive test repetitions. It is estimated that the lost test bed time due to errors detected too late is up to 40%.
It can also be assumed that many errors are not detected at all, which in the worst case can lead to momentous wrong decisions being made in the development process due to the incorrect evaluation of a technology.
For economic reasons, it is therefore crucial to avoid such events by early and reliable detection of faulty sensors and measurement deviations. In general, the earlier a fault is detected, the sooner troubleshooting measures can be initiated to continue the test productively. The complexity of modern combustion engines with a high number of manipulated and controlled variables usually requires an equally extensive sensor system in the test, which makes manual monitoring of the measurement data quality increasingly difficult. Together, these factors mean that the need for automated fault diagnosis is continuously growing.

Diagnostic software adaptable to new test vehicles and test tasks

One of the main challenges in diagnosing faults on engine test benches is that they are often characterized by frequent changes of the test vehicle, i.e. they are usually systems that cannot always be clearly described technically in advance. Under these general conditions, the requirement for adaptability and flexible usability must be particularly emphasized.
It must be possible to adapt the diagnostic software quickly to new test vehicles and test tasks. For reasons of user-friendliness, however, the configuration effort, i.e. the number of parameters that have to be set, must be as low as possible.
The LEC MCheck software enables comprehensive and reliable monitoring of the measurement online and directly at the test stand. The quality of the measured data is displayed in the form of a traffic light, which means that the monitoring of a complex system can be reduced to a single primary parameter to be observed. Of course, detailed and extensive detailed results are available for more in-depth analyses, as will be shown below. In the case of insufficient measurement data quality, the software is able to identify faulty sensors, so that the correct measures, such as the replacement or recalibration of sensors or measuring devices, can be initiated in a timely manner.

Modular software architecture for any type of test bed

LEC MCheck thus not only supports and relieves the test bed personnel, but also has the potential to significantly contribute to increasing the overall efficiency of an engine test bed. In addition, a continuous self-validation of the monitoring quality is performed. The user therefore has detailed information available at any time about how well individual sensors or measuring devices can be monitored.
This function prevents incorrect parameterization of the diagnostic system and also ensures that no wrong expectations are created. The modular concept, as well as the combination of data-based and physical methods, allow a fast and efficient adaptation to special test tasks or new test vehicles. Although LEC MCheck was developed specifically for engine test beds, the modular and expandable software architecture also allows it to be used on other test bed systems.

Downloads

PDF Product Folder

PDF Product Folder – version to print on one sheet

Software

LEC MCheck

LEC MCheck Key Features

Get in touch

LEC MCheck in practice

Downloads