Questions & Answers

Hi,I am working in TrendMiner Machine Learning Hub (MLHub) and need to save/download files generated by a Python notebook, such as Parquet or ZIP files.I found that files created under /home/jovyan do not persist after a kernel restart. I also tried downloading a generated ZIP file through an IPython FileLink, but this did not work.Small files can be downloaded from notebook output using a base64 data URI, but a 10 MB generated file does not render the download link. What is the supported way to download larger Python-generated files (such as Parquet/ZIP files) from MLHub, or persist them outside the ephemeral /home/jovyan filesystem?What is the supported method for either:downloading notebook-generated files to my local computer, or storing them in a persistent MLHub location that survives kernel restarts?Thank you.

What is the best way to dynamically identify data shifts like this in Trendminer?  Control chart calculations would be ideal, but I don’t think that’s possible. 

Hi team Trendminer,I am encountering a bug with the Collapsed Filters. In the Trendhub View below I have 4 signals over 2 trends on which a filter is applied. The filter is based on a simple Value Based Search, which is returning the results I expect. This is confirmed in both the Trendview and the Context View.  As soon as I toggle ‘Collapsed Filters’, the filtering overlay disappears and no collapsed plotting is applied: With every toggle an error is spawn in the browser console: ​​​

I was able to follow the instructions on another thread on how to create an aggregation tag for an integration over 24 hours of a binary tag for line running or not and it will give me the downtime over the last 24 hours at any specific point. Is there a way to lock in a time so that on any given day at 7 am, it will show me the downtime over the last 24 hours to the previous day at 7 am? We have a few metrics like that that we report out on over the 7 am - 7 am window and that feature would be very helpful.

Is there a way to clear the Recent Items tile on the homepage ? 

Hello TrendMiner forum.We are currently onboarding new LDAP users and have encountered an issue when attempting to share dashboards with edit permissions.IssueWhen using the Share Dashboard functionality, we are unable to find newly onboarded LDAP users in the user selection list. However, the list contains many older users, including users who are no longer present in the system.ObservationsNew LDAP users do not appear when searching for users to share a dashboard with. Existing users also cannot find the new users. New users cannot find other new users either. For example, User A and User B were both recently created and neither user appears in the other's share list. Surprisingly, we can see numerous old users that are no longer listed in ConfigHub > Security > User Overview as either LDAP or Local users. We are not currently using groups for dashboard permissions. As a test, I created a group and added one of the new users along with myself, but we still could not see each ot

Hi TrendMiner Team,I'm working on a bagging process where the weighing scale provides a continuous analog weight tag. I have created context items for each completed bag and can analyze them in Event Analytics.My goal is to build a DashHub dashboard that displays one weight value for each completed bag over time (i.e., a trend of final bag weights, not the continuous analog filling curve).However, I've noticed that: ContextHub only allows me to display context metadata (Start Date, End Date, Duration, Description, etc.). Event Analytics allows me to calculate statistics such as Maximum or Last Weight for each bag event, but I cannot find a way to display those Event Analytics results on a DashHub dashboard as a live trend. My questions are: Is it possible to display one calculated value (e.g., Maximum or Last Weight) per context item/event as a time trend on a DashHub dashboard? Can Event Analytics results be linked directly to DashHub? If not, what is the recommended TrendMiner

We are using TrendMiner 2025.R4.0-12.A monitor automatically creates a Context Item for every completed bag.Event Analytics correctly calculates the Maximum Weight for each bag.Is there a supported way to automatically populate a numeric Context Item Field (e.g. Bag Weight) with that Event Analytics calculation so it becomes available in ContextHub and DashHub?If not, is there another supported method to expose one calculated value per monitor-generated event on a DashHub dashboard?

I would like to use a specific date in a formula for a condition; for example: if (date < exchange_date, constant1 * tag1, constant 2 * tag1)I tried this with TM_day, TM_month tags, but they don't give the exact months/dates when using in a formular (perhaps because they are converted from strings?)

I am using a conditional statement to calculate the U-value of heat exchanger. If conditions are not met, I would like to return a nan rather than 0. How can I do this?if(    and((Q_warm < 13), (Q_warm > 11)),    (Q/(3.64*LMTD)),    0)

Hi everyone,I am trying to build a highly accurate, custom binary (0/1) signal to track a specific cooking phase on a production line. The standard state tags stay active too long, and I need an out-of-the-box solution using TrendMiner's formula tool or search capabilities to handle misaligned PLC signals.The Challenge: Start Condition: A state tag transitions to >= 3000 and < 7000. This triggers the start of the cooking phase. The Process: While in this state, a cooling/discharge valve cycles ON (1) and OFF (0). The Problem: The state tag continues to sit at 5000+ long after the physical process is done. A point-by-point IF statement formula fails because if the valve isn't open the exact second the state hits 3000, the formula evaluates to 0 and misses the batch. The Goal: I want a calculated tag or capsule that jumps to 1 when the state hits 3000, stays at 1 through the process, and drops to 0 the exact moment the valve goes from 1 -> 0 (falling edge), completely retu

I am trying to understand predictive monitoring concepts in TrendMiner using a cooking process as an example.Suppose historical data shows that longer cooking cycles are often preceded by slower temperature rise rates, lower steam pressure, or abnormal valve behavior.What TrendMiner features would you recommend for detecting these conditions early and warning operators before the cooking cycle exceeds its normal duration?I am particularly interested in understanding the progression from trend analysis and Value Based Search toward predictive monitoring and would appreciate examples of how others have implemented this in process manufacturing.

Suppose a cooker gradually becomes less efficient over several months.For example: Temperature rise becomes slower Steam usage increases Cooking duration increases What is the best TrendMiner approach to detect these long-term performance changes before they become operational problems?Would dashboards, monitors, fingerprints, or another method be most suitable?

I found a cooking cycle with a very specific temperature profile that I would like to use as a reference.Rather than manually reviewing months of data, I would like to identify all historical cooking cycles that exhibit a similar process pattern.What is the recommended TrendMiner workflow for searching historical data using a known process pattern as the reference?Are there best practices for improving search accuracy when dealing with cooking processes?

Our cooking process follows a sequence of heating, cooking, cooling, and transfer.The total cycle duration occasionally increases, but it is not immediately clear which stage is responsible.What is the best way in TrendMiner to automatically break the cycle into stages and determine where most delays are occurring?For example: Heating delay Extended cooking period Cooling delay Transfer delay Has anyone implemented a similar workflow?

I have several cooking cycles that represent stable and efficient operation.I would like to use these cycles as a reference and automatically compare future cooking cycles against them.What is the best way to create a representative cooking profile or fingerprint in TrendMiner and detect when future cycles begin to deviate from normal behavior?Can this be automated for continuous monitoring?

I have identified multiple cooking cycles and noticed that some take significantly longer than others.Possible contributing factors may include: Steam pressure Steam valve position Product level Initial product temperature Cooling duration What would be the recommended TrendMiner workflow to compare shorter versus longer cooking cycles and determine which process variables are most strongly associated with increased cooking duration?Are there specific TrendMiner features that simplify this type of root cause analysis?

I am working with a cooking process where the sequence generally consists of: Steam valve opens Temperature rises Product cooks Cooling starts Transfer valve opens Product transfer completes I would like to automatically identify every complete cooking cycle over several months of historical data.Currently I can identify individual events using tag values, but I am interested in understanding the most robust TrendMiner approach for detecting complete cooking cycles and creating a reusable context for future analysis.Would you use Value Based Search, ContextHub, Event Analytics, or another method?How would you structure the logic to reliably identify the start and end of each cooking cycle?

We have a transfer pump that moves product between process stages.Historically the transfer operation takes approximately 8–10 minutes. Over the last few months operators have reported that some transfers appear to be taking longer, although the issue is not consistent enough to trigger alarms.Available historian data includes: Pump running status Motor current Flow rate Product level Discharge pressure I would like to investigate whether the pump performance is gradually degrading over time and identify which process variables would be most useful for monitoring this.What TrendMiner workflow would you recommend to: Detect gradual performance deterioration. Compare current pump performance against historical operation. Build an early warning system before transfer times become unacceptable.

I am using a Value Based Search on a recipe step tag with the condition StepNo >= 3000 AND StepNo <= 5000 to identify cooking periods. Step 7000 is the transfer phase and should not be included. However, the search results appear to extend beyond the cooking phase and display step-change markers for Step 7000. Is there a way to force the result boundaries to stop when the tag leaves the 3000–5000 range, or should I create a formula tag such as CookingPhase = 1 when StepNo is between 3000 and 5000 and search on that instead?

A heat exchanger is used to heat product before entering production. Over time: Product outlet temperature decreases. Steam consumption increases. Production throughput drops. Questions: How can I use TrendMiner to quantify heat exchanger efficiency over time? How can I determine when cleaning is required? Can I automatically detect fouling before it impacts production?

Hello Community,We are experiencing a strange issue on one of our trend charts and are hoping someone can point us in the right directionWe are trending a PLC tag for a pump/valve that strictly has two operating states: 1 (Idle)and 2 (Running). The data flows from the PLC to our historian, and we are visualizing it inside TrendMiner.The Issue:Instead of seeing a clean block or step chart alternating between 1 and 2, the trend line is occasionally jumping or spiking to values like 5 or 6. None of these values exist in the PLC logic for this device

I am trying to set up a critical control point (CCP) real-time monitor for a sterilization process on a fermenter line.My goal is simple: I want an alarm to trigger on our operator dashboard if the tank temperature stays above 120 outside of a strict 19-minute to 25-minute window. (i.e., if it is under-sterilized or over-sterilized).There are no fields, dropdowns, or advanced criteria boxes on this screen that allow me to specify the 19-minute or 25-minute duration limits. It only allows me to select the Type (Information/Alarm) and type a text description. i want to check if it the time window exceeded and also didnt got up to the time window frame 

Hi,I have a question regarding the monitor and the context items being created.I created a monitor based on a saved search for the fermenter CCP sterilization process. The monitor is configured to create context items when the search conditions are met.My concern is about how the context items are updated over time.For example, if I initially create and save the ContextHub view using a time range from June 4 to June 9, and later new sterilization events occur on June 10, June 11, and beyond: Will the monitor automatically create new context items for these future events? Will the ContextHub dashboard automatically display these newly created context items? Or is the ContextHub view limited to the original time range that was selected when it was created? If I want operators to continuously see new CCP violations (e.g., sterilization periods that are too short or too long), do I need to configure the ContextHub view in a specific way? I just want to make sure the monitor continues