272 lines
11 KiB
Python
272 lines
11 KiB
Python
import json
|
||
import threading
|
||
import time
|
||
import pandas
|
||
import math
|
||
import csv
|
||
import numpy
|
||
from common import clibs
|
||
|
||
|
||
def initialization(path, hr, data_dirs, data_files, interval, curves, w2t):
|
||
def check_files():
|
||
nonlocal interval
|
||
if interval == "":
|
||
interval = 300
|
||
elif interval.isdigit():
|
||
if int(interval) < 300:
|
||
w2t(f"输入时间间隔 {interval} < 300，使用默认时间间隔 300s ......\n", "orange")
|
||
interval = 300
|
||
else:
|
||
interval = int(interval)
|
||
else:
|
||
w2t(f"{interval} 不是有效的输入，时间间隔必须是一个大于 300 的正整数！\n", "red", "NotIntegerError")
|
||
|
||
if len(curves) == 0:
|
||
w2t("未查询到需要记录数据的曲线，至少选择一个！\n", "red", "CurveNameError")
|
||
|
||
if len(data_dirs) != 0 or len(data_files) != 1:
|
||
w2t("初始路径下不允许有文件夹，且初始路径下只能存在一个工程文件 —— *.zip，确认后重新运行！\n", "red", "InitFileError")
|
||
|
||
if not data_files[0].endswith(".zip"):
|
||
w2t(f"{data_files[0]} 不是一个有效的工程文件，需确认！\n", "red", "ProjectFileError")
|
||
|
||
return data_files[0], interval
|
||
|
||
def get_configs():
|
||
robot_type, records = None, None
|
||
try:
|
||
msg_id, state = hr.execution("controller.get_params")
|
||
records = hr.get_from_id(msg_id, state)
|
||
except Exception:
|
||
w2t("网络不可达，需要先连接xCore！\n", "red", "NetworkError")
|
||
for record in records:
|
||
if "请求发送成功" not in record[0]:
|
||
robot_type = eval(record[0])["data"]["robot_type"]
|
||
server_file = f"/home/luoshi/bin/controller/robot_cfg/{robot_type}/{robot_type}.cfg"
|
||
local_file = path + f"/{robot_type}.cfg"
|
||
clibs.c_pd.pull_file_from_server(server_file, local_file)
|
||
|
||
try:
|
||
with open(local_file, mode="r", encoding="utf-8") as f_config:
|
||
configs = json.load(f_config)
|
||
except Exception as Err:
|
||
clibs.insert_logdb("ERROR", "current", f"get_config: 无法打开 {local_file}，获取配置文件参数错误 {Err}")
|
||
w2t(f"无法打开 {local_file}\n", color="red", desc="OpenFileError")
|
||
|
||
# 最大角速度，额定电流，减速比，额定转速
|
||
version = configs["VERSION"]
|
||
m_avs = configs["MOTION"]["JOINT_MAX_SPEED"]
|
||
m_rts = configs["MOTOR"]["RATED_TORQUE"] # 电机额定转矩rt for rated torque
|
||
m_tcs = [1, 1, 1, 1, 1, 1] # 电机转矩常数，tc for torque constant
|
||
m_rcs = []
|
||
for i in range(len(m_tcs)):
|
||
m_rcs.append(m_rts[i] / m_tcs[i]) # 电机额定电流，rc for rated current
|
||
clibs.insert_logdb("INFO", "do_brake", f"get_configs: 机型文件版本 {robot_type}_{version}")
|
||
clibs.insert_logdb("INFO", "do_brake", f"get_configs: 各关节角速度 {m_avs}")
|
||
clibs.insert_logdb("INFO", "do_brake", f"get_configs: 各关节额定电流 {m_rcs}")
|
||
return m_avs, m_rcs
|
||
|
||
prj_file, interval = check_files()
|
||
avs, rcs = get_configs()
|
||
params = {
|
||
"prj_file": prj_file,
|
||
"interval": interval,
|
||
"avs": avs,
|
||
"rcs": rcs,
|
||
}
|
||
return params
|
||
|
||
|
||
def change_curve_state(hr, curves, stat_1, stat_2):
|
||
display_pdo_params = [{"name": name, "channel": chl} for name in curves for chl in range(6)]
|
||
hr.execution("diagnosis.open", open=stat_1, display_open=stat_2, overrun=True, turn_area=True, delay_motion=False)
|
||
hr.execution("diagnosis.set_params", display_pdo_params=display_pdo_params, frequency=50, version="1.4.1")
|
||
|
||
|
||
def run_rl(path, params, curves, hr, md, w2t):
|
||
prj_file, interval = params["prj_file"], params["interval"]
|
||
# 1. 关闭诊断曲线，触发软急停，并解除，目的是让可能正在运行着的机器停下来，切手动模式并下电
|
||
change_curve_state(hr, curves, False, False)
|
||
md.r_soft_estop(0)
|
||
md.r_soft_estop(1)
|
||
md.r_clear_alarm()
|
||
md.write_act(False)
|
||
time.sleep(1) # 让曲线彻底关闭
|
||
|
||
# 2. reload工程后，pp2main，并且自动模式和上电
|
||
prj_name = ".".join(prj_file.split("/")[-1].split(".")[:-1])
|
||
prj_path = f"{prj_name}/_build/{prj_name}.prj"
|
||
hr.execution("overview.reload", prj_path=prj_path, tasks=["factory"])
|
||
hr.execution("rl_task.pp_to_main", tasks=["factory"])
|
||
hr.execution("state.switch_auto")
|
||
hr.execution("state.switch_motor_on")
|
||
|
||
# 3. 开始运行程序
|
||
hr.execution("rl_task.set_run_params", loop_mode=True, override=1.0)
|
||
hr.execution("rl_task.run", tasks=["factory"])
|
||
t_start = time.time()
|
||
while True:
|
||
if md.read_ready_to_go() == 1:
|
||
md.write_act(True)
|
||
break
|
||
else:
|
||
if (time.time() - t_start) > 3:
|
||
w2t("3s 内未收到机器人的运行信号，需要确认RL程序编写正确并正常执行...\n", "red", "ReadySignalTimeoutError")
|
||
else:
|
||
time.sleep(1)
|
||
|
||
# 4. 获取初始数据，周期时间，首次的各轴平均电流值，打开诊断曲线，并执行采集
|
||
time.sleep(10) # 等待 RL 程序中 scenario_time 初始化
|
||
t_start = time.time()
|
||
while True:
|
||
scenario_time = float(f"{float(md.read_scenario_time()):.2f}")
|
||
if scenario_time != 0:
|
||
w2t(f"耐久工程的周期时间：{scenario_time}s | 单轮次执行时间：{scenario_time+interval}\n")
|
||
break
|
||
else:
|
||
time.sleep(1)
|
||
if (time.time() - t_start) > 300:
|
||
w2t(f"300s 内未收到耐久工程的周期时间，需要确认RL程序和工具通信交互是否正常执行...\n", "red", "GetScenarioTimeError")
|
||
|
||
# 6. 准备数据保存文件
|
||
for curve in curves:
|
||
with open(f"{path}/{curve}.csv", mode="a+", newline="") as f_csv:
|
||
titles = [f"{curve}_{i}" for i in range(6)]
|
||
titles.insert(0, "time")
|
||
csv_writer = csv.writer(f_csv)
|
||
csv_writer.writerow(titles)
|
||
|
||
# 7. 开始采集
|
||
count = 0
|
||
while clibs.running:
|
||
if not clibs.running:
|
||
w2t("后台数据清零完成，现在可以重新运行之前停止的程序。\n", "red")
|
||
exit()
|
||
|
||
this_time = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime(time.time()))
|
||
next_time = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime(time.time()+scenario_time+interval+1))
|
||
w2t(f"[{this_time}] 当前次数：{count:09d} | 预计下次数据更新时间：{next_time}\n", "#008B8B")
|
||
count += 1
|
||
# 固定间隔，更新一次数据，打开曲线，获取周期内电流，关闭曲线
|
||
time.sleep(interval)
|
||
change_curve_state(hr, curves, True, True)
|
||
time.sleep(scenario_time)
|
||
end_time = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime(time.time()))
|
||
start_time = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime(time.time()-scenario_time))
|
||
change_curve_state(hr, curves, False, False)
|
||
# 保留数据并处理输出
|
||
gen_results(params, curves, start_time, end_time, w2t)
|
||
|
||
|
||
def gen_results(params, curves, start_time, end_time, w2t):
|
||
try:
|
||
clibs.lock.acquire(True)
|
||
clibs.cursor.execute(f"select content from logs where time between '{start_time}' and '{end_time}' and content like '%diagnosis.result%' order by id asc")
|
||
records = clibs.cursor.fetchall()
|
||
finally:
|
||
clibs.lock.release()
|
||
data_proc(records, params, curves, w2t)
|
||
|
||
|
||
def data_proc(records, params, curves, w2t):
|
||
for curve in curves:
|
||
if curve == "device_servo_trq_feedback":
|
||
# proc_device_servo_trq_feedback(records, params, w2t)
|
||
t = threading.Thread(target=proc_device_servo_trq_feedback, args=(records, params, w2t))
|
||
t.daemon = True
|
||
t.start()
|
||
elif curve == "hw_joint_vel_feedback":
|
||
# proc_hw_joint_vel_feedback(records, params, w2t)
|
||
t = threading.Thread(target=proc_hw_joint_vel_feedback, args=(records, params, w2t))
|
||
t.daemon = True
|
||
t.start()
|
||
|
||
|
||
def proc_device_servo_trq_feedback(records, params, w2t):
|
||
d_trq, rcs, results = [[], [], [], [], [], []], params["rcs"], [time.strftime("%Y-%m-%d %H:%M:%S", time.localtime(time.time()))]
|
||
for record in records:
|
||
data = eval(record[0])["data"]
|
||
for item in data:
|
||
d_item = reversed(item["value"])
|
||
for axis in range(6):
|
||
if item.get("channel", None) == axis and item.get("name", None) == "device_servo_trq_feedback":
|
||
d_trq[axis].extend(d_item)
|
||
|
||
for axis in range(6):
|
||
df = pandas.DataFrame.from_dict({"device_servo_trq_feedback": d_trq[axis]})
|
||
_ = math.sqrt(df.apply(lambda x: numpy.power((rcs[axis] * float(x.iloc[0]) / 1000), 2)).sum() / len(df))
|
||
results.append(_)
|
||
|
||
path = "/".join(params["prj_file"].split("/")[:-1])
|
||
with open(f"{path}/device_servo_trq_feedback.csv", mode="a+", newline="") as f_csv:
|
||
csv_writer = csv.writer(f_csv)
|
||
csv_writer.writerow(results)
|
||
|
||
|
||
def proc_hw_joint_vel_feedback(records, params, w2t):
|
||
d_trq, rcs, results = [[], [], [], [], [], []], params["rcs"], [time.strftime("%Y-%m-%d %H:%M:%S", time.localtime(time.time()))]
|
||
for record in records:
|
||
data = eval(record[0])["data"]
|
||
for item in data:
|
||
d_item = reversed(item["value"])
|
||
for axis in range(6):
|
||
if item.get("channel", None) == axis and item.get("name", None) == "hw_joint_vel_feedback":
|
||
d_trq[axis].extend(d_item)
|
||
|
||
for axis in range(6):
|
||
df = pandas.DataFrame.from_dict({"hw_joint_vel_feedback": d_trq[axis]})
|
||
_ = df.max().iloc[0]
|
||
results.append(_)
|
||
|
||
path = "/".join(params["prj_file"].split("/")[:-1])
|
||
with open(f"{path}/hw_joint_vel_feedback.csv", mode="a+", newline="") as f_csv:
|
||
csv_writer = csv.writer(f_csv)
|
||
csv_writer.writerow(results)
|
||
|
||
|
||
def detect_db_size():
|
||
@clibs.db_lock
|
||
def release_memory():
|
||
line_number = 20000
|
||
leftover = 4000 # 200s
|
||
clibs.cursor.execute("select count(id) from logs")
|
||
len_records = clibs.cursor.fetchone()[0]
|
||
if len_records > line_number:
|
||
del_num = len_records - leftover + 1
|
||
clibs.cursor.execute(f"delete from logs where id < {del_num}")
|
||
clibs.cursor.execute(f"update logs set id=(id-{del_num-1}) where id > {del_num-1}")
|
||
clibs.cursor.execute(f"update sqlite_sequence set seq = {leftover+1} WHERE name = 'logs' ")
|
||
clibs.cursor.execute("vacuum")
|
||
|
||
while True:
|
||
release_memory()
|
||
time.sleep(1)
|
||
|
||
|
||
def main():
|
||
t = threading.Thread(target=detect_db_size)
|
||
t.daemon = True
|
||
t.start()
|
||
|
||
path = clibs.data_dd["path"]
|
||
interval = clibs.data_dd["interval"].strip()
|
||
curves = clibs.data_dd["curves"]
|
||
hr = clibs.c_hr
|
||
md = clibs.c_md
|
||
w2t = clibs.w2t
|
||
|
||
data_dirs, data_files = clibs.traversal_files(path, w2t)
|
||
params = initialization(path, hr, data_dirs, data_files, interval, curves, w2t)
|
||
prj_file = params["prj_file"]
|
||
clibs.c_pd.push_prj_to_server(prj_file)
|
||
try:
|
||
run_rl(path, params, curves, hr, md, w2t)
|
||
except Exception as Err:
|
||
w2t(f"工厂耐久程序执行过程中出现异常，{Err}\n", "red")
|
||
change_curve_state(hr, curves, False, False)
|
||
|
||
|
||
if __name__ == "__main__":
|
||
main()
|