Mika Ruusunen

Predictive information on plant diseases could help to reduce and optimize the usage of pesticides in agriculture. This research presents classification procedures with linear discriminant analysis to predict three possible severity levels of net blotch in spring barley in Finland. The weather data utilized for classification included mathematical transformations, namely features of outdoor temperature and air humidity with calculated dew point temperature values. Historical field observations of net blotch density were utilized as a target class for the identification of classifiers. The performance of classifiers was analyzed in sliding data windows of two weeks with selected, cumulative, summed feature values. According to classification results from 36 yearly data sets, the prediction of net blotch occurrence in spring barley in Finland can be considered as a linearly separable classification task. Furthermore, this can be achieved with linear discriminant analysis by combining the output probabilities of separate binary classifiers identified for each severity level of net blotch disease. In this case, perfect classification with a resolution of three different net blotch severity levels was achieved during the first 50 days from the beginning of the growing season. This strongly suggests that real-time classification based on a few weather variables measured on a daily basis can be applied to estimate the severity of net blotch in advance. This allows application of the principles of integrated pest management (IPM) and usage of pesticides only when there is a proven need.

The aim of wastewater treatment plants (WWTPs) is to clean wastewater before it is discharged into the environment. Real-time monitoring and control will become more essential as the regulations for effluent discharges are likely to become stricter in the future. Model-based soft sensors provide a promising solution for estimating important process variables such as chemical oxygen demand (COD) and help in predicting the performance of WWTPs. This paper explores the possibility of using interpretable model structures for monitoring the influent and predicting the effluent of paper mill WWTPs by systematically finding the best model parameters using an exhaustive algorithm. Experimentation was conducted with regression models such as multiple linear regression (MLR) and partial least squares regression (PLSR), as well as LASSO regression with a nonlinear scaling function to account for nonlinearities. Some autoregressive time series models were also built. The results showed decent modelling accuracy when tested with test data acquired from a wastewater treatment process. The most notable test results included the autoregressive model with exogenous inputs for influent COD (correlation 0.89, mean absolute percentage error 8.1%) and a PLSR model for effluent COD prediction (correlation 0.77, mean absolute percentage error 7.6%) with 20 h prediction horizon. The results show that these models are accurate enough for real-time monitoring and prediction in an industrial WWTP.

Well-performing control loops have an integral role in efficient and sustainable industrial production. Control performance monitoring (CPM) tools are necessary to establish further process optimization and preventive maintenance. Data-driven, model-free control performance monitoring approaches are studied in this research by comparing the performance of nine CPM methods in an industrially relevant process simulation. The robustness of some of the methods is considered with varying fault intensities. The methods are demonstrated on a simulator which represents a validated state-space model of a supercritical carbon dioxide fluid extraction process. The simulator is constructed with a single-input single-output unit controller for part of the process and a combination of relevant faults in the industry are introduced into the simulation. Of the demonstrated methods, Kullback–Leibler divergence, Euclidean distance, histogram intersection, and Overall Controller Efficiency performed the best in the first simulation case and could identify all the simulated fault scenarios. In the second case, integral-based methods Integral Squared Error and Integral of Time-weighted Absolute Error had the most robust performance with different fault intensities. The results highlight the applicability and robustness of some model-free methods and construct a solid foundation in the application of CPM in industrial processes.

Barkhausen noise (BN) measurements are commonly used for surface characterisation. However, often there is also a need to verify the sub-surface region because detrimental tensile stresses may be present after different manufacturing steps. Especially in a grinding burn, the surface stress may be compressive, but it changes quickly into tensile stress below the surface. The aim of this study was to find out whether regular surface-sensitive BN measurement is also sensitive to the stresses below the surface caused by grinding burns. More specifically, the aim was to study the relationship between BN features and sub-surface stresses and to identify a model that estimates sub-surface stresses. Real samples were collected from an actual process. The samples were cylindrical samples manufactured from commercial alloyed AISI/SAE L6 steel that was through-hardened prior to grinding. Barkhausen noise measurements were carried out for 42 grinding burn locations followed by X-ray diffraction-based residual stress surface measurements and residual stress depth profiles. Depth information was obtained through step-by-step electrolytic removal of thin layers. The stress profiles were pre-processed through interpolation and averaged stress was calculated as a function of depth below the surface. Correlation analysis was carried out to evaluate the relationships between BN features and stress at different depths and among BN features. The main outcome of the analysis was that BN measurement is dominated by the sub-surface tensile stresses rather than the compressive stress at the surface. It was also noticed that BN features form two groups, corresponding to average Barkhausen activity and magnetising field strength leading to maximum Barkhausen activity. Models for stress at different steps were identified systematically. The performance of the models for sub-surface stresses was reasonable with R² values of around 0.85 and root mean squared error (RMSE) values of around 95 MPa. Based on the results, it is concluded that BN measurement provides information about sub-surface stresses and that stress can be evaluated through straightforward modelling, allowing fast detection of grinding burns.

The performance of meteorological data-based methods to forecast plant diseases strongly depends on temporal weather information. In this paper, a data analysis procedure is presented for finding the optimal starting time for forecasting net blotch density in spring barley based on meteorological data. For this purpose, changes in the information content of typically measured weather variables were systemically quantified in sliding time windows and with additionally generated mathematical transformations, namely with features. Signal-to-noise statistics were applied in a novel way as a metric for identifying the optimal starting time instance and the most important features to successfully distinguish between two net blotch densities during springtime itself. According to the results, the information content of meteorological data used in classifying between nine years with and four years without net blotch reached its maximum in Finnish weather conditions on the 41st day from the beginning of the growing season. Specifically, utilising weather data at 41–55 days from the beginning of the growing season maximises successful forecasting potential of net blotch density. It also seems that this time instance enables a linear classification task with a selected feature subset, since the averages of the metrics in two data groups differ statistically with a minimum 68% confidence level for nine days in a 14-day time window.