This paper focuses on the damage identification problem of pile wharves’ foundation piles and conducts a quantitative analysis study based on the BP neural network. Traditional damage localization methods are influenced by subjective human factors during the identification process, and their effectiveness is limited when only first-order modal data is available. To overcome these issues, a reasonable finite element model of the pile wharf is established, and training and testing are conducted using simulated data from ABAQUS and actual vibration signal data, to construct a damage quantitative analysis model based on the BP neural network. The results prove that the damage indicators ULSC and δFC under the first and second modal conditions are suitable for the neural network of damage quantification in single damage scenarios, and they can accurately determine the degree of damage to damaged elements.