I have the same problem here. I'm using a HPMA115S0 with an ESP8266 and found out that the sensor don't respond in general if you are doing a deep sleep with the ESP. If the ESP wakes up, the sensor does not respond on commands till you remove the 5 volts of the HPMA for a short time to make an "hard reset" of the sensor.
I think there is a problem with the ttl-level on the UART-RX-line of the HPMA. Because if you're pulling the rx-line of the sensor manually to ground (maby the ESP is doing this if he goes to deep sleep or wakes up), the sensor does not respond after that.
I spend very much time in this phenomenon. I used different pins on the ESP for software UART and used the hardware UART, problem still exists.
I assume that the ESP pulls all his pins on ground level for too long. That means that the stop bit (high-level), which the sensor wants after 8 bits are transmitted, is not present. This leads to strange behavior of the sensor.
I don't have a digital storage scope to see the level on rx-line, so this is only a presumption. But I don't have this problem, if the ESP idles with an sleep-command.
Maby you are also doing a deepsleep with your ESP which results in the same problem?