Strategy for Water Level Control Based on Prediction Control Feedback Model

First, this article uses the fuzzy comprehensive evaluation algorithm to consider the interests and costs of all parties to construct a comprehensive satisfaction score as the objective function, with the safety needs of residents around the Great Lakes for water levels as constraints, and the optimal water level is obtained through nonlinear programming. Then, this article uses the Predictive-Control-Feedback (PCF) concept to construct a water level control management model. The predictive model is mainly built on the differential equation model of the flow rate change per second for each lake. The LSTM model predicts the per second precipitation, evaporation, and runoff. The input and output flow rates of the river are calculated based on the Manning formula. The PCF model dynamically minimizes the sum of squared errors between the predictive model and the target by using grid search to find the optimal control flow rate. In addition to the differential equations, the predictive model also includes a recursive predictive correction control term, which provides feedback to the current predictive model based on the deviation between the actual value and the predicted value from the previous period. Meanwhile, we have tested the sensitivity of this algorithm to dam outflow and environmental changes, and the result shows that the algorithm is stable for both types of sensitivity tests.