Repository logo
 

Research Publications (Engineering and Built Environment)

Permanent URI for this collectionhttp://ir-dev.dut.ac.za/handle/10321/215

Browse

Subject: search.filters.subject.Artificial neural networks

Search Results

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Thermocouple signal conditioning using augmented device tables and table look-up neural networks, with validation in J-Thermocouples
    (IEEE, 2022-01-25) Maseko, Moses L.; Agee, John T.; Davidson, Innocent
    The relatively high accuracy, large measurement range, and durability of thermocouple devices make these devices to probably be the most-widely used temperature measuring devices in industrial applications. The ability of thermocouples to sense temperature is derived from the generation of thermoelectric voltages arising due to temperature differences between the hot and cold junctions of the thermocouple. Thermocouple temperature measurement processes suffer from inaccuracies arising from both the unwanted or undetected variations in the cold junction temperature of the thermocouple, and nonlinearities in the generated thermoelectric voltage. This paper presents an enhancement of thermocouple temperature measurement using a combination of augmented thermocouple tables generated from thermocouple polynomial functions, look-up MLP neural networks trained to accept the thermocouple output voltage, and the cold or reference junction temperature measurements: to produce improved hot-junction temperature outputs. Experimental validation of the current approach for a J thermocouple, using data from augmented device tables, reproduced the measured temperature values with a worst-case error of 0.0094%.
  • Thumbnail Image
    Item
    Comparison of two data-driven modelling techniques for long-term streamflow prediction using limited datasets
    (SCIELO, 2015-09) Oyebode, Oluwaseun Kunle; Adeyemo, Josiah; Otieno, Fredrick Alfred O.
    This paper presents an investigation into the efficacy of two data-driven modelling techniques in predicting streamflow response to local meteorological variables on a long-term basis and under limited availability of datasets. Genetic programming (GP), an evolutionary algorithm approach and differential evolution (DE)-trained artificial neural networks (ANNs) were applied for flow prediction in the upper uMkhomazi River, South Africa. Historical records of streamflow, rainfall and temperature for a 19-year period (1994-2012) were used for model design, and also in the selection of predictor variables into the input vector space of the model. In both approaches, individual monthly predictive models were developed for each month of the year using a one-year lead time. The performances of the predictive models were evaluated using three standard model evaluation criteria, namely mean absolute percentage error (MAPE), root mean-square error (RMSE) and coefficient of determination (R2). Results showed better predictive performance by the GP models (MAPE: 3.64%; RMSE: 0.52: R2: 0.99) during the validation phase when compared to the ANNs (MAPE: 93.99%; RMSE: 11.17; R2: 0.35). Generally, the GP models were found to be superior to the ANNs, as they showed better performance based on the three evaluation measures, and were found capable of giving a good representation of non-linear hydro-meteorological variations despite the use of minimal datasets.