为解决多源数据下物资供应链数据分配不均匀或存在特征混淆的问题，提出基于多源数据的物资供应链数据均衡处理方法。建立物资供应链数据分布空间，设定非边界和边界区域，标定区域中心，分别计算不同特征数据与中心点间的距离，采用边缘混合采样算法预先查找非均衡数据，根据数据距离特征判定其所属的边界区域。将非均衡数据看作是供应链集群中的工作节点，当集群中工作节点数量发生变化时，通过流量计算的方式求得每次循环执行节点产生的负载，将负载值转换为执行线程分数，并通过分数对比得到不均衡数据，并计算物资供应链数据内存和CPU资源耗用占比阈值，建立并行任务线程，利用调度器将工作节点中的执行线程转移到下一节点来达到负载均衡目的，从而实现数据均衡处理。试验结果表明：所提方法对电力物资数据均衡处理的响应时间短、数据吞吐量大、处理效果好、稳定性强，具有较好的实际应用价值。

In order to solve the problem of uneven data distribution or confusion of characteristics in material supply chain under multi-source data, a data balance processing method of material supply chain based on multi-source data was proposed. The data distribution space of material supply chain was established, non-boundary and boundary areas were set, the center of the area was calibrated, the distances between different characteristic data and the center point were calculated respectively, the unbalanced data was searched in advance by using the edge mixed sampling algorithm, and the boundary area to which it belongs was determined according to the data distance characteristics. The unbalanced data was regarded as the working nodes in the supply chain cluster. When the number of working nodes in the cluster changes, the load generated by each cycle execution node was obtained by flow calculation, and the load value was converted into the execution thread score. The unbalanced data was obtained by comparing the scores, the threshold of data memory and CPU resource consumption in the material supply chain was calculated, and the parallel task thread was established. The scheduler was used to transfer the execution thread in the working node to the next node to achieve the purpose of load balancing. The experimental results show that the proposed method has the advantages of short response time, large data throughput, good processing effect and strong stability, and has good practical application value.