def __init__()

in lab/app/windfarm.py [0:0]

• 21 lines of code
• 7 McCabe index (conditional complexity)

    def __init__(self, simulator, mqtt_host, mqtt_port):
        if simulator is None:
            raise Exception("You need to pass the simulator as argument")

        self.simulator = simulator
        self.n_turbines = simulator.get_num_turbines()

        self.mqtt_host = mqtt_host
        self.mqtt_port = mqtt_port

        ## launch edge agent clients
        self.edge_agents = [EdgeAgentClient('/tmp/agent%d' % i) for i in range(self.n_turbines)]
        self.model_meta = [{'model_name':None} for i in range(self.n_turbines)]
        self.ota_devices = []

        # we need to load the statistics computed in the data prep notebook
        # these statistics will be used to compute normalize the input
        self.raw_std = np.load('statistics/raw_std.npy')
        self.mean = np.load('statistics/mean.npy')
        self.std = np.load('statistics/std.npy')
        # then we load the thresholds computed in the training notebook
        # for more info, take a look on the Notebook #2
        self.thresholds = np.load('statistics/thresholds.npy')
        
        # configurations to format the time based data for the anomaly detection model
        # If you change these parameters you need to retrain your model with the new parameters        
        self.INTERVAL = 5 # seconds
        self.TIME_STEPS = 20 * self.INTERVAL # 50ms -> seg: 50ms * 20
        self.STEP = 10
        
        # these are the features used in this application
        self.feature_ids = [8,9,10,7,  22, 5, 6] # qX,qy,qz,qw  ,wind_seed_rps, rps, voltage        
        self.n_features = 6 # roll, pitch, yaw, wind_speed, rotor_speed, voltage
        self.running = False # running status
        
        # minimal buffer length for denoising. We need to accumulate some sample before denoising
        self.min_num_samples = 500