In view of the connection requirements of safe disposal and reuse of industrial plants in the process of urban renewal, this paper constructs a collaborative technical framework of protective demolition and structural reconstruction of industrial plants. A unified coding, coordinate alignment and state mapping mechanism is established around building surveying and mapping data, laser point clouds, close-range images, historical drawings, component damage records, demolition sequence records, material recovery data, reconstruction structural parameters and construction monitoring sequences, forming a continuous information chain covering evaluation before demolition, control during demolition and reconstruction after demolition. On this basis, combined with component retention identification, demolition disturbance propagation, reconstruction constraint matching and feedback writeback, the protective demolition decision model, reconstruction unit configuration model and construction feedback linkage mechanism are constructed, so that the geometric attributes, connection relationships, damage states, reuse values and dynamic observation results of components can be continuously updated in the same state space. The experiment is based on four typical industrial plant samples. The results show that the accuracy of component retention recognition reaches 93.4%, the vibration control standard rate reaches 91.8%, the adaptive rate of reconstructed components reaches 90.6%, and the average collaborative response time is controlled at 1.9 s. This method can better coordinate the relationship between demolition safety, component retention and structural reconstruction, and is suitable for industrial plant function conversion, space restoration and renovation scenarios, and can provide a continuous basis for site review, path verification and control implementation.
Povzetek: Za potrebe zaščitnega razstavljanja in strukturne rekonstrukcije industrijskih obratov v procesu urbane prenove ta članek vzpostavlja usklajen tehnični okvir, ki omogoča enotno kodiranje večvirnih podatkov, preslikavo stanj ter povratni vnos in neprekinjeno posodabljanje podatkov. Eksperimentalni rezultati kažejo, da je natančnost prepoznavanja ohranjenih komponent dosegla 93,4 %, stopnja skladnosti nadzora vibracij 91,8 %, stopnja prilagodljivosti rekonstruiranih komponent 90,6 %, povprečni odzivni čas pa 1,9 s.
