This research proposes the optimisation model of propaganda resource allocation in the field of public security. This model can quantitatively evaluate the effect of public security propaganda. It solves the problem of maximising the effect of public security propaganda in the condition of fixed resources, and the optimisation model was verified by using the data of fraud cases. Define the propaganda cost and the propaganda objects, and set minimising the loss of social victimisation as the optimisation objective of the model. The optimisation model was derived from the average loss and the possibility of victimisation of different propaganda objects. The Bayesian formula is used to transform the statistical information of a specific population. Simplify the objective function, and construct the Lagrange function. Calculate the weight matrix and parameter vector, utilise KTT condition or MATLAB to solve the nonlinear optimization equation. By using the audience rating data and the data of the age of the victims and the amount of the victims cheated in D city, the proportion of the investment needed in different types of media was obtained according to the objective data, and the propaganda resource allocation scheme which could directly guide the practice was obtained. The optimisation model of resource allocation has strong operability and the conclusions drawn from this model are precise and clear. This model is based on objective data with strong operability, which is simple to apply, convenient to calculate.