public override async Task ProcessItem(dynamic eventData)
{
// Ensure this is a correctly-formatted event for ML; ignore it otherwise
if (eventData == null || eventData.deviceid == null || eventData.cycle == null ||
eventData.sensor9 == null || eventData.sensor11 == null || eventData.sensor14 == null || eventData.sensor15 == null)
{
return;
}
// The experiment theoretically supports multiple inputs at once,
// even though we only get one value at a time, so the request
// requires an array of inputs
MLRequest mlRequest = new MLRequest(ML_REQUEST_COLUMNS, new string[, ]
{
{
// The id is required to be numeric, so we hash the actual device id
eventData.deviceid.ToString().GetHashCode().ToString(),
// The remaining entries are string representations of the numeric values
eventData.cycle.ToString(),
eventData.sensor9.ToString(),
eventData.sensor11.ToString(),
eventData.sensor14.ToString(),
eventData.sensor15.ToString()
}
}
);
HttpClient http = new HttpClient();
http.DefaultRequestHeaders.Authorization = new AuthenticationHeaderValue("Bearer", _configurationProvider.GetConfigurationSettingValue("MLApiKey"));
http.BaseAddress = new Uri(_configurationProvider.GetConfigurationSettingValue("MLApiUrl") + ML_ENDPOINT);
HttpResponseMessage response = await http.PostAsJsonAsync("", mlRequest);
if (response.IsSuccessStatusCode)
{
MLResponse result = JsonConvert.DeserializeObject <MLResponse>(await response.Content.ReadAsStringAsync());
RulTableEntity entry = new RulTableEntity
{
PartitionKey = eventData.deviceid.ToString(),
RowKey = eventData.cycle.ToString(),
// Extract the single relevant RUL value from the JSON output
Rul = result.Results["data"].value.Values[0, RUL_COLUMN],
// Since the simulator might replay data, ensure we can overwrite table values
ETag = "*"
};
// We don't need a data model to represent the result of this operation,
// so we use a stub table/model convertor
await AzureTableStorageHelper.DoTableInsertOrReplaceAsync <object, RulTableEntity>(entry, (RulTableEntity e) => null,
_configurationProvider.GetConfigurationSettingValue("eventHub.StorageConnectionString"),
_configurationProvider.GetConfigurationSettingValue("MLResultTableName"));
}
else
{
throw new Exception(string.Format("The ML request failed with status code: {0}", response.StatusCode));
}
}
public override async Task ProcessItem(dynamic eventData)
{
// Ensure this is a correctly-formatted event for ML; ignore it otherwise
if (eventData == null || eventData.deviceid == null || eventData.cycle == null ||
eventData.sensor9 == null || eventData.sensor11 == null || eventData.sensor14 == null || eventData.sensor15 == null)
{
return;
}
// The experiment theoretically supports multiple inputs at once,
// even though we only get one value at a time, so the request
// requires an array of inputs
MLRequest mlRequest = new MLRequest(ML_REQUEST_COLUMNS, new string[,]
{
{
// The id is required to be numeric, so we hash the actual device id
eventData.deviceid.ToString().GetHashCode().ToString(),
// The remaining entries are string representations of the numeric values
eventData.cycle.ToString(),
eventData.sensor9.ToString(),
eventData.sensor11.ToString(),
eventData.sensor14.ToString(),
eventData.sensor15.ToString()
}
}
);
HttpClient http = new HttpClient();
http.DefaultRequestHeaders.Authorization = new AuthenticationHeaderValue("Bearer", _configurationProvider.GetConfigurationSettingValue("MLApiKey"));
http.BaseAddress = new Uri(_configurationProvider.GetConfigurationSettingValue("MLApiUrl") + ML_ENDPOINT);
HttpResponseMessage response = await http.PostAsJsonAsync("", mlRequest);
if (response.IsSuccessStatusCode)
{
MLResponse result = JsonConvert.DeserializeObject<MLResponse>(await response.Content.ReadAsStringAsync());
RulTableEntity entry = new RulTableEntity
{
PartitionKey = eventData.deviceid.ToString(),
RowKey = eventData.cycle.ToString(),
// Extract the single relevant RUL value from the JSON output
Rul = result.Results["data"].value.Values[0, RUL_COLUMN],
// Since the simulator might replay data, ensure we can overwrite table values
ETag = "*"
};
// We don't need a data model to represent the result of this operation,
// so we use a stub table/model convertor
await AzureTableStorageHelper.DoTableInsertOrReplaceAsync<object, RulTableEntity>(entry, (RulTableEntity e) => null,
_configurationProvider.GetConfigurationSettingValue("eventHub.StorageConnectionString"),
_configurationProvider.GetConfigurationSettingValue("MLResultTableName"));
}
else
{
throw new Exception(string.Format("The ML request failed with status code: {0}", response.StatusCode));
}
}
public override async Task ProcessItem(dynamic eventData)
{
// Ensure this is a correctly-formatted event for ML; ignore it otherwise
if (eventData == null || eventData.deviceid == null || eventData.cycle == null ||
eventData.sensor9 == null || eventData.sensor11 == null || eventData.sensor14 == null || eventData.sensor15 == null)
{
return;
}
string result = await _mlServiceInvoker.GetRULAsync(
// The id is required to be numeric, so we hash the actual device id
eventData.deviceid.ToString().GetHashCode().ToString(),
// The remaining entries are string representations of the numeric values
eventData.cycle.ToString(),
eventData.sensor9.ToString(),
eventData.sensor11.ToString(),
eventData.sensor14.ToString(),
eventData.sensor15.ToString()
);
Trace.TraceInformation($"RUL Result: {result}");
RulTableEntity entry = new RulTableEntity
{
PartitionKey = eventData.deviceid.ToString(),
RowKey = eventData.cycle.ToString(),
// Extract the single relevant RUL value from the JSON output
Rul = result,
// Since the simulator might replay data, ensure we can overwrite table values
ETag = "*"
};
// We don't need a data model to represent the result of this operation,
// so we use a stub table/model convertor
await AzureTableStorageHelper.DoTableInsertOrReplaceAsync <object, RulTableEntity>(entry, (RulTableEntity e) => null,
_configurationProvider.GetConfigurationSettingValue("eventHub.StorageConnectionString"),
_configurationProvider.GetConfigurationSettingValue("MLResultTableName"));
}