Tianjin is a typical industrialized city of 13.64 million people, and the urbanization rate is 85.49%. The risk of heavy metals in the soils of the typical agricultural land around Tianjin poses a significant challenge to the sustainability of the ecosystem’s health and human health. Different heavy metals in different land-use types in Tianjin have all accumulated in the soils, and the vegetable base had the highest total of accumulated heavy metals. This study took the surface soil of farmland Xiqing District—the main vegetable and crop area in Tianjin—as the research object, and the concentrations of eight heavy metals were analyzed. The geo-accumulation index (I_geo), principal component analysis (PCA), absolute principal component score-multiple linear regression (APCS–MLR), positive definite matrix factorization (PMF), and health risk assessment model were used to evaluate the degree, sources, and health risks (to adults and children) of heavy metal pollution. This study compares the APCS–MLR model with the PMF model. The results showed that Cd and Hg pollution were the most severe among the eight heavy metals in agricultural soil, with the average values exceeding the background by 151.9% and 324.1%, respectively. About 15% of the sites were at moderate to severe pollution levels. The PMF model can better analyze the sources of heavy metals in the study area, showing that the main sources of heavy metal pollution include natural source, mixed source of agriculture and transportation, coal combustion source, and pesticide source. The total carcinogenic risk index (TCR) of natural source is the highest, with Cr being the main contributor to maximum total non-carcinogenic risk indices (HI) and TCR for children; Hg contributes the most to HI in the coal combustion source, while Cu and Zn contributes most in the mixed source of agriculture and transportation.

In order to solve the problem of complex coupling characteristics between multivariate load sequences and the difficulty in accurate multiple load forecasting for integrated renewable energy systems (IRESs), which include low-carbon emission renewable energy sources, in this paper, the TCN-FECAM-Informer multivariate load forecasting model is proposed. First, the maximum information coefficient (MIC) is used to correlate the multivariate loads with the weather factors to filter the appropriate features. Then, effective information of the screened features is extracted and the frequency sequence is constructed using the frequency-enhanced channel attention mechanism (FECAM)-improved temporal convolutional network (TCN). Finally, the processed feature sequences are sent to the Informer network for multivariate load forecasting. Experiments are conducted with measured load data from the IRES of Arizona State University, and the experimental results show that the TCN and FECAM can greatly improve the multivariate load prediction accuracy and, at the same time, demonstrate the superiority of the Informer network, which is dominated by the attentional mechanism, compared with recurrent neural networks in multivariate load prediction.