diff --git "a/AscendPytorch\346\250\241\345\236\213\346\216\250\347\220\206\344\274\227\346\231\272FAQ.md" "b/AscendPytorch\346\250\241\345\236\213\346\216\250\347\220\206\344\274\227\346\231\272FAQ.md"
index 55b39e877ea24e80795411ec306821bbba1b3b6f..5629481dbc79bb34d947a5bfa8e7f611c0647ac3 100644
--- "a/AscendPytorch\346\250\241\345\236\213\346\216\250\347\220\206\344\274\227\346\231\272FAQ.md"
+++ "b/AscendPytorch\346\250\241\345\236\213\346\216\250\347\220\206\344\274\227\346\231\272FAQ.md"
@@ -128,6 +128,7 @@ x = F.adaptive_avg_pool2d(input, output_size=bin_size)替换为x = adaptive_avg_
     修改/etc/sudoers将Defaults env_reset改成Defaults !env_reset
     修改/etc/bash.bashrc添加alias sudo='sudo env PATH=$PATH LD_LIBRARY_PATH=$LD_LIBRARY_PATH'
     ```
+
 ### FAQ14、推理性能不达标，profiling显示TransData算子耗时，参考如下方案优化
 （1）修改five_2_four.py优化方法  
   在环境变量env.sh中export install_path=/usr/local/Ascend/ascend-toolkit/latest路径下查找five_2_four.py文件，路径一般为
@@ -177,6 +178,49 @@ atc转换时指定输出节点类型为float16
 atc --framework=5 --model=./ICNet.onnx --output=ICNet_bs1 --out_nodes="Resize_317:0" --output_type=FP16 --input_format=NCHW --input_shape="actual_input_1: 1,3,1024,2048" --log=debug --soc_version=Ascend310
 ```
 
+### FAQ15、onnx转om模型报错atc命令ERROR问题解决
+* 现象描述
+    ```
+    ATC run failed，please check the detail log. try 'atc --help'
+    E19999: Inter Error!
+            Unknown error occurred,please check the log.
+    ```
+    1. 设置环境变量
+   ```
+   export install_path=/usr/local/Ascend/ascend-toolkit/latest   
+   export PATH=/usr/local/python3.7.5/bin:${install_path}/atc/ccec_compiler/bin:${install_path}/atc/bin:$PATH   
+   export PYTHONPATH=${install_path}/atc/python/site-packages:$PYTHONPATH   
+   export LD_LIBRARY_PATH=${install_path}/atc/lib64:${install_path}/acllib/lib64:$LD_LIBRARY_PATH   
+   export ASCEND_OPP_PATH=${install_path}/opp  
+  ```
+2. 更新最新的推理包run包
+
+3. 打印host日志
+   ```
+   export ASCEND_SLOG_PRINT_TO_STDOUT=1
+   [WARNING] TBE(3112,atc.bin):2021-05-25-15:20:33.329.360 [image_ops.cc:2146][OP_PROTO] ResizeNearestInferShape:2146 OpName:[Resize_140] "Get 
+   constValue failed of [sizes]"
+   [ERROR] TBE(3112,atc.bin):2021-05-25-15:20:33.329.371 [image_ops.cc:2084][OP_PROTO] CalculateSizeOut:2084 OpName:[Resize_140] "length of scale_out 
+    after erase must be equal to 2"
+   [ERROR] TBE(3112,atc.bin):2021-05-25-15:20:33.329.376 [image_ops.cc:2155][OP_PROTO] ResizeNearestInferShape:2155 OpName:[Resize_140] "calculate size 
+   out failed."
+   [ERROR] GE(3112,atc.bin):2021-05-25-15:20:33.329.391 [op_desc.cc:1345]3112 CallInferFunc: ErrorNo: -1(failed) [COMP][PRE_OPT]Resize_140 call infer 
+   func. ret: 4294967295
+   [ERROR] GE(3112,atc.bin):2021-05-25-15:20:33.329.397 [shape_refiner.cc:766]3112 InferShapeAndType: ErrorNo: -1(failed) [COMP][PRE_OPT]Resize_140 call 
+   infer function failed.
+   ```
+   得出的结论为：onnx不支持 constValuse 需要进行优化转换  
+   优化转换采用onnx-simplifier 工具进行转换  
+   安装onnx-simplifier  
+   pip3 install onnx-simplifier   
+   简化onnx模型：  
+   python3 -m onnxsim ./hrnet_w18.onnx ./hrnet_w18_1.onnx --input-shape "16,3,224,224"  
+   转换完成再执行如下命令  
+   ```
+   atc --framework=5 --model=./hrnet_w18_1.onnx --input_format=NCHW --input_shape="image:16,3,224,224" --output=hrnet_bs16 --log=debug --
+   soc_version=Ascend310
+   ```  
+   onnx转om成功。
 
 ## [2.2 NPU模型精度调试常见问题FAQ](#22-NPU模型精度调试常见问题FAQ)
 
diff --git a/mobilenet_w18_bs32.rar b/mobilenet_w18_bs32.rar
new file mode 100644
index 0000000000000000000000000000000000000000..854a3f8d55ac0ac511d5a0b449538c3b2cb99a27
Binary files /dev/null and b/mobilenet_w18_bs32.rar differ