FIELD: water treatment.
SUBSTANCE: detection of anomalies during the operation of water treatment plants. The method contains stages in which data is obtained from sensors reflecting the operating state of the plant, and additional data, and these data are processed in a system for collecting and processing, which carries out a training stage, during which the system calculates the parameters of the probability distribution, taking into account the specified data, and the specified algorithm produces a pointer reflecting the general state of the plant in the form of a probabilistic law; stage of applying the algorithm, in which the values read by the sensors in real time are entered, the probability density values are calculated for all sensors and, depending on the result of the calculation of this probability density value, if this probability density value is low, it is concluded that the sensors provide values different from the values they provided during the training phase and then indicate an anomaly in the operation of the installation; wherein this probability law is a simplified version with independent Gaussian distributions or a version with a multivariate Gaussian distribution.
EFFECT: increasing the battery safety.
2 cl, 1 dwg
