Research On The Key Techniques Of Locat- -ing The Source Of Pollution Emergency Occurred In River Channel

River pollution emergencies occurred frequently in recent years which have made significant impact to people’s life. Current source locating methods could not achieve satisfactory results since their slow responding speed and low detecting accuracy. Accordingly, an algorithm framework with higher suitability to different usage scenarios is proposed. This paper focuses on the key techniques of pollution source inversion algorithm, on-line parameter optimization, source locating results sampling analysis and so on. The proposed method has been tested with simulations and experiments to verify its validation and performances.The main contents and innovative points are summarized as follows.(1) A river pollution source inversion algorithm is proposed in this paper in order to solve the problem existing in pollution location identifying. It converts source locating problem into a Bayesian evaluation problem. With the observation data of monitoring section and the pollution expansion model, a likelihood function with pollution source information is built to obtain the posteriori probability density function based on Bayesian theorem. The possible spatial position, emission time and other unknown parameters can be calculated by sampling and analyzing the posteriori probability density function.(2) An on-line dynamic parameters optimization method is proposed in order to improve the real-time performance and the accuracy of the optimization algorithm. The pollution source simulation and downstream pollution density prediction are performed by utilizing the initial section data instead of the actual pollution source. By analyzing the differences between the prediction values and the measured values, the model parameters can be dynamically corrected. The calculation results denoted that the parameters can be calibrated and updated online with actual data to achieve the optimal performance.(3) As the sampling results can not easily represent the real situation accurately when analyzing result features, a self-adaptive Metropolis-Hastings sampling method is proposed to solve this problem. The proposed method uses self-adaptive calculation to obtain standard deviation of the proposal function, thus it solves the non-convergent problem caused by the instability of acceptable probability. The calculation results of a real case demonstrate that with the help of posteriori probability, it can obtain the parameters such as emission volume, spatial position and leaking time stably, quickly and accurately.(4) Two experiments were carried out in wave tank to simulate the river pollution event. After injecting tracer into the tank, we took the measured concentration data in the monitoring points as inversion data, and calculated the probability distribution of the pollution source. Thoroughly, the practicability of the proposed pollution locating method along with its online parameter calibration is verified to be effective by comparing with real values.The research work of this paper improved the technical accumulation of the pollution source locating method. The results of this work will be integrated into the water quality early warning system in the near future. And the whole system is designed to assist the related administrations to get hold of the pollution source information and make decisions in time.