In response to teachers’ insufficient knowledge management ability, this study intends to optimize the important parameters of support vector machines using lightning search algorithms to obtain improved algorithms. Research constructs a new evaluation model for teacher knowledge management ability based on improved algorithms. The performance comparison analysis of the improved algorithm proposed in the study showed that the accuracy and area under the PR curve of the algorithm were 93.47% and 0.8, superior to the comparison algorithm. In empirical analysis, using this evaluation model to improve teachers’ knowledge management ability resulted in teacher satisfaction scores of 94 points and student satisfaction scores of 96 points, both better than before the improvement. The above results indicated that the improved algorithm and teacher knowledge management ability evaluation model had good performance. Therefore, this model can be used to improve teachers’ knowledge management ability, thereby promoting the overall quality development of school teachers.