Predictive maintenance is one of the pillar of Industry 4.0, especially for industries integrating machineries subjected to continuous failures which sensibly impact on operations, up to completely stop processes and production. In this perspective, the paper addresses the problem of continuous monitoring of industrial filter press aimed at the early detection blockages/damages of the membranes before avalanche phenomena arise, to prevent the overall failure of the machinery and a prolonged downtime of the plant. The monitoring system includes more than 150 battery-less UHF RFID temperature sensors properly positioned on the inflating pipes of the filter press membranes and a long-range reading architecture comprising two readers and multiple antenna arrays. The system is controlled by an embedded pc unit capable to stream measured data to the SCADA system of the factory via OPC-UA protocol. The paper illustrates the principle of operations, the selection of the components, the deployment of the system and the early operative results.