Water temperature affects biochemical reactions in aquatic organisms and regulates the maximum dissolved oxygen concentratio n, so evaluating and monitoring this parameter is necessary. A computational model is presented for assessing water temperature in intensive culture ponds of white shrimp Litopenaeus vannamei by studying three fundamental factors : the average temperature, amplitude variation and duration of temperature changes. Through a fuzzy reasoning system, each factor is evaluated using rules, obtaining an indicator of the impact of temperature on the shrimp habitat. The results were compared against the NSF and CCME water quality indices, the most used indices. It shows better behaviour in the daily temperature evaluation , which more appropriately penalizes abrupt temperature changes, which is not considered by the other indices.
The first step in the proposal computational model is the signal preprocessing. The raw data has null samples due to acquisition device errors. The cubic splines are used to reconstruct the signal. Then, a smoothing process is applied to the new signal to remove high-frequency noise.
The smoothed signa l is split into 96 segments; every segment corresponds to a w hole day from 00:00 to 23:45 hours. The average daily temperature for every segment is calculated as the first fundamental factor temperature average.
Peaks and valleys are calculated in every signal segment to estimate the second and third evaluation factors. Those peak-valley pairs with an absolute temperature difference greater than a threshold °C are selected. The average temperature and the time are calculated as the amplitude variation and duration of temperature changes, respectively.
Once the three fundamental factors are estimated, they are used to feed an inference fuzzy system (FIS) whose output corresponds to the evaluation temperature of water.
The results of the proposal computational model are shown in Figure 1.