PrecisionLivestock Optimization Model for Enhancing Sustainable Animal Welfare Management

Abstract

Raising livestock is increasingly under pressure to balance production objectives with humane animal welfare and environmental sustainability. Conventional precision livestock farming (PLF) implementation of such systems measure feeding, thermal comfort and health parameters as separate subsystems and therefore has a restricted ability to solve trade-offs between competing welfare targets and resources. The authors present the PrecisionLivestock Optimization Model (PLOM), a single decision model that combines multiple sensors and their associated welfare parameters into a single composite optimization objective for feed allocation, water management, thermal comfort control, and early detection of welfare risks. PLOM treats the problem of welfare enhancement as a constrained multi-objective optimization problem, optimising animal-centred scores of welfare, while considering operational costs and environmental impacts, and solves the problem using an adaptive weighted-objective search strategy that adapts the priorities of subsystems as new evidence from the sensors becomes available. The model is validated using a simulation based case study of a mixed dairy and poultry enterprise which is compared with a conventional rule based PLF benchmarking using ten iterations of optimization and four key subsystems. The results simulated suggest that PLOM has a higher value of the composite welfare index than the baseline and has seen gains in subsystem efficiency of between approximately fifteen and twenty-six percentage points, greatest for the subsystem relating to thermal comfort regulation and health monitoring. The results indicate that a single optimization target for welfare indicators instead of multiple targets for each indicator separately provides a realistic route towards animal-centred precision livestock management for sustainable production. The paper also provides comments on considerations for implementation, data needs, and limitations for farm deployment in real-world contexts, and directions for the validation of the framework based on data gathered from the field for various species and farming situations.