Structural Approach To Design Sensor Networks for Fault Diagnosis

Abstract

Key faults significantly affect the normal operation of the process originating risk conditions. These failures should be identified even in the presence of missing measurements or outliers. In this work a new strategy to design sensor networks, which are able to resolve a set of key faults when sensors fail, is presented. The procedure deals with failure isolation using the Fault Resolution Degree concept. This is incorporated as a constraint of the minimum-cost design formulation, and the resulting optimization problem is solved using MILP codes. The strategy only uses low uncertainty data that are readily available at the process design stage. Application results of the methodology to case studies extracted from the literature are presented and compared with those provided by other existing techniques.

Sensor Location for Enhancing Fault Diagnosis

“…The second one takes into account that all the iv ’s should be different. This condition is formulated using the concept of the NNV …”

“…This condition is formulated using the concept of the NNV. 6 The solution of the proposed optimization problems is carried out using an improved LTTS. Given that fault observability is a necessary condition for fault detection and that fault resolution has the same role for its isolation, FOR constraints are incorporated to the SNUP because they can be posed as linear inequalities.…”

“…To test condition 2, three vectors are defined in terms of iv j and iv k as follows where v 1 contains all the measured variables of iv j not included in iv k , in contrast v 2 involves the set of observations of iv k not contained in iv j , and v 3 has the common elements. It has been demonstrated that faults j and k are isolatable if at least two of the three vectors ( v1 , v2 , and v3 ) are non-null, that is, if the number of non-null vectors that arise from the conjunction between iv j and iv k , denoted as NNV jk ( q ), is greater than 2. It has been demonstrated that the previous inequality can be replaced by the following set of constraints …”

“…It has proved that FOR are satisfied if the subsequent conditions are fulfilled where rv j = ( a j ∧ q ) is the resolution vector ( rv ) of the j th failure, and rv i , j = 1 means that the i th variable is measured and its value changes when this fault takes place. It should be remarked that FOR conditions can be easily checked using a set of [ J + J ( J – 1)/2] linear inequalities.…”

“…Comprehensive reviews about the variety of methods proposed to locate instruments with fault diagnosis purposes have been presented. − The instrumentation selection strategies for processes to be monitored using PCA satisfy at least the observability and resolution of a predetermined set of faults. In this way, the detection and isolation of those failures are partially ensured. , It has been suggested to select sensors at the design stage by applying directed-graph-based procedures or a technique that assumes the fault probability values for the instruments and the process .…”