The question is how much and what data do I need to prepare for the data warehouse so that the AI \u200b\u200bcan start working? In this mode, the furnaces will perform 6,066 cycles and process 3,033 tons of parts per year.<\/p>\n The decisive factor is not the total number of records, but mainly:<\/p>\n The number of records in the table below refers to individual completed and identified batches or cycles, not individual second-by-second temperature records.<\/p>\n The best approach is to create a model on the initial set and then observe the learning curve: the accuracy of the model is evaluated at 50, 100, 200, 400 and other records. The curve will show whether additional data still improves the result, or saturation is already occurring.<\/p>\n The model that will be used for such a quenching plant will be a product model. We will therefore collect data related to the product, containing everything related to the material and cycle, quality data. It will be good if we create a process family, where we will also associate similar processes and materials, and then later include similar materials and similar products in this family.<\/p>\n There must be a clear link from the data set: RFQ<\/strong> and quotation, material and material condition, basic geometry and weights, number of pieces and proposed number of pieces in the batch, fixture and layout of parts on the fixture, recipe, required cycle parameters (temperature, pressure, gas type), AI<\/strong> proposed SPT (Standard Process Time)<\/strong>\u00a0cycle time, AI calculated consumption per cycle, prediction of CHD, NHD, hardness and deformation results, quotation costs, risks of deviations and rework.<\/p>\n With 300-500 quality cases, you can usually:<\/p>\n For a broader production model covering more technologies, I would aim for at least: 1,000\u20133,000 full cycles, divided by process families.<\/p>\n The selected model should first be limited to, for example:<\/p>\n So the practical goal is not to obtain hundreds of records for each product. A reasonable goal is to create a common model from a few thousand C45 cycles and obtain approximately 30\u201350 well-described reference cycles for each regular product.<\/p>\n If the process is stable and known:<\/p>\n After one year, the data warehouse could contain, for example: 3,000\u20135,000 usable C45 cycles, data from all three furnaces, different dimensions, different preparations, different casts, different oil conditions, resulting hardnesses and deformations, costs.<\/p>\n The model does not relearn the entire process. It only finds out, for example: This product has an average hardness of 1.5 HRC lower<\/strong> than the general C45 model under the same conditions. Or: When filled above 85%, the deformation<\/strong> of this product increases by 0.04 mm.<\/p>\n Of course, in the first case I also have to store hardness measurements in the data warehouse, in the second case deformation measurements.<\/p>\n However, at the beginning we said that such a hardening plant has 100 to 200 products. For such a heat treatment plant, this means from 200*30 to 200*100 cycles, or data tied to a quenching cycle, i.e. from 6,000 to 20,000 structured data records. This is a huge amount of data and the AI \u200b\u200blayer<\/strong> administrator alone is not enough. It has to be handled differently. Since all ERP<\/strong> systems are based on relational tables, the above data should not be a problem to prepare if the ERP system knows it.<\/strong> But we are back to ERP quality.<\/strong><\/p>\n To get started, we need to make sure that the ERP system<\/strong> gives us the basic data structure for export to the AI \u200b\u200blaye<\/strong>r. However, this historical backfill must contain the data that we actually need to obtain from the ERP<\/strong> \u2013 see infographic. The abbreviation ETL<\/strong> stands for Extract-Transform-Load. After processing the data in the middleware, the data is moved to the data warehouse.<\/p>\n Once the system is launched, new records are automatically added. So if we import 2 years for this heat treatment plant model, we have 12,000 cycle records. The initial data is prepared by an ERP consultant, data engineer, integrator and technologist who will explain the meaning of the fields. The fact that we do not have process data for cycles at this point \u2013 temperature, pressure, flow rates, dew point, Cp, is not crucial.<\/strong> The crucial thing is that the rest is there. And here we are back to Cycles<\/strong>. If we do not have cycle management in ERP<\/strong>, if we do not have a deterministic cost calculation model in ERP,<\/strong> we do not have data for AI either<\/strong>. In this case, I assume that it is simply not worth it at all. Even AI<\/strong> will not solve the missing data for us. Even if ?<\/p>\n ======================================================================================================= AI<\/strong>:\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0A<\/strong>rtificial\u00a0I<\/strong>ntelligence![\"\"](\"https:\/\/intelligent-kalam.80-240-27-133.plesk.page\/wp-content\/uploads\/2026\/08\/chatgpt-image-29-7-2026-23-32-34.png\")
<\/a><\/p>\n\n
<\/a><\/p>\n\n
<\/a><\/p>\n\n
<\/a><\/p>\n<\/a><\/p>\n
\nPou\u017eit\u00e9 zkratky:<\/strong><\/em><\/span><\/p>\n
\n<\/em>API:\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0<\/strong>Application Programming\u00a0I<\/strong>nterface: definovan\u00e9 rozhran\u00ed a soubor pravidel, podle kter\u00fdch si dva syst\u00e9my p\u0159ed\u00e1vaj\u00ed data a p\u0159\u00edkazy. Webov\u00e9 API m\u016f\u017ee b\u00fdt dostupn\u00e9 prost\u0159ednictv\u00edm jedn\u00e9 nebo v\u00edce URL adres, ozna\u010dovan\u00fdch jako endpoints.
\n<\/strong><\/em>JSON: \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0<\/strong>JavaScript\u00a0O<\/strong>bject\u00a0N<\/strong>otation: textov\u00fd form\u00e1t pro strukturovan\u00fd p\u0159enos dat. Vznikl v prost\u0159ed\u00ed JavaScriptu, ale dnes je nez\u00e1visl\u00fd na programovac\u00edm jazyce.
\n<\/em>Middleware: \u00a0\u00a0<\/strong>integra\u010dn\u00ed vrstva, kter\u00e1 propojuje syst\u00e9my, transformuje data, \u0159\u00edd\u00ed komunikaci, zabezpe\u010den\u00ed a auditn\u00ed z\u00e1znamy.
\n<\/em>AI model: \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0<\/strong>matematick\u00fd nebo statistick\u00fd model, kter\u00fd prov\u00e1d\u00ed klasifikaci, predikci, detekci odchylek nebo doporu\u010den\u00ed.
\n<\/em>OPC UA<\/strong>: \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 Open Platform Communications nebo OPC Unified Architecture
\n<\/em>AI vrstva: \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0<\/strong>\u0159e\u0161en\u00ed sestaven\u00e9 ze standardn\u00edch platforem, konektor\u016f, pravidel a individu\u00e1ln\u011b nakonfigurovan\u00fdch nebo vyvinut\u00fdch model\u016f.
\n<\/em>Document AI:\u00a0<\/strong>Technologie vyu\u017e\u00edvan\u00e1 k\u00a0extrakci vyu\u017e\u00edv\u00e1 k\u00a0extrakci informac\u00ed<\/a>\u00a0z ti\u0161t\u011bn\u00fdch i\u00a0digit\u00e1ln\u00edch dokument\u016f<\/a>, jako jsou obr\u00e1zky, texty, znaky
\n<\/em>DMS:\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0 D<\/strong>ocument\u00a0M<\/strong>anagement\u00a0S<\/strong>ystem
\n<\/em>FQ<\/strong>: \u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0R<\/strong>equest\u00a0F<\/strong>or\u00a0Q<\/strong>uotation
\n<\/em>APQP:<\/strong>\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0A<\/strong>dvanced\u00a0P<\/strong>roduct\u00a0Q<\/strong>uality\u00a0P<\/strong>lanning
\n<\/em>PPAP:<\/strong>\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0P<\/strong>roduction\u00a0P<\/strong>art\u00a0A<\/strong>pproval\u00a0P<\/strong>rocess
\n<\/em>ERP:<\/strong>\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0\u00a0E<\/strong>nterprise\u00a0R<\/strong>esource\u00a0P<\/strong>lanning<\/em><\/span><\/p>\n