Method for automatically processing excavated material on a conveyor equipped with a hyperspectral imager

Abstract

The present invention relates to a method for automatically processing excavated material in the form of aggregate on a conveyor provided with a hyperspectral imager (10), characterised in that it comprises means for comparing, in real time, the imaged region (8) of the flow of excavated material with a hyperspectral base (15) of signatures characteristic of undesirable constituents, said means controlling a means (4) for deflecting the flow towards a secondary container (3) if undesirable constituents are detected in the imaged region (8) and / or a means (4) for deflecting the flow towards another secondary container (3) if desirable constituents are detected in the imaged region (8).

Description

Scope of the invention

[0001] The present invention relates to the field of recycling and recovery of excavated material and rubble from public works sites, as well as aggregates from quarries (non-secondary raw material).

[0002] In France, construction and demolition waste represents a resource of nearly 250 million tons per year, ten times more than household waste. In 2015, only 40% of this waste was recycled and recovered, while the law on energy and environmental transition sets a recovery rate of 70% by 2020.

[0003] While large companies in the sector are organizing themselves, construction tradespeople, subject to the same regulations, are struggling to find solutions. The majority of their construction waste ends up buried without being recycled.

[0004] Rapid characterization of different forms of waste in real time is a complex operation given the large volume of excavated material, and it is difficult to objectively and rigorously determine the presence of: Inert waste (stones, soil and earthmoving materials, plaster, ceramics, unmixed demolition materials, etc.): Inert waste is stable over time; it does not decompose, burn, or produce any chemical, physical, or biological reactions during storage. It comes from the end-of-life of natural products (stones, soil, sand) or manufactured products (concrete, ceramics, bricks, glass) and from uncontaminated sediment or excavated soil. Non-hazardous waste or industrial waste (wood, plastics, metals, hardware, etc.) is waste produced by various types of industries (including construction), commerce, and services. It is not hazardous or toxic, but it cannot be considered inert. This waste is either composed of a single material (untreated wood, various metals, plaster, bitumen, etc.)This includes composite materials, products associated with plaster (such as insulation composites), fibrous materials (with particular attention to asbestos), treated glass, and plastics. Packaging waste, as well as contaminated sediments or excavated soil, also fall into this category. Non-inert waste requiring specific treatment and which cannot be mixed with inert waste includes paints, wood treated with heavy metal oxides, friable asbestos, hydrocarbons, and other materials. This type of waste contains toxic substances and requires specific treatment during disposal. Examples include treated wood, paints, solvents and varnishes, chemical treatment products (antioxidants, fungicides, abrasives, detergents, etc.), as well as asbestos, electrical / electronic building products, and heavily contaminated sediments or excavated soil.

[0005] The RECYBETON project produced a report "Theme 1 Selective sorting technology for crushed concrete aggregates" Author: C. DEBOFFE - NEO ECO R / 15 / RECY / 027 LC / 13 / RECY / 31 September 2015 presenting the general principles of sorting construction and demolition waste.

[0006] ADEME also presented a report entitled "STATE OF THE ART OF TECHNOLOGIES"

[0007] "IDENTIFICATION AND SORTING OF WASTE" presenting different sorting technologies (https: / / www.ademe.fr / sites / default / files / assets / documents / 87753_rapport-ajeurope-tec hnologies-de-tri.pdf). State of the art

[0008] Prior art patent application US2019299255 describes a system and method for sorting scrap metal particles. A mobile conveyor containing scrap metal particles is imaged using a vision system to create a vision image corresponding to a synchronized location on the conveyor, and is detected using a detection system to create a detection matrix corresponding to the synchronized location. This prior art detection system includes at least one array of analog proximity sensors. A control system analyzes the vision image as a cell vision matrix and generates a vision vector containing vision data from the vision matrix for the particle.This control system analyzes the detection matrix and generates a detection data vector containing detection data from the detection matrix for the particle. The control system classifies the particle into one of at least two material classifications based on the vision data vector and the detection data vector. Disadvantages of prior art

[0009] Prior art solutions are not suitable for high-throughput automated sorting of rubble for on-site recycling of construction and demolition waste. They are suitable for identifying the presence of one or more compounds based on re-emission, generally through fluorescence at specific wavelengths, for the purposes of analysis and overall classification of materials moving along the conveyor. Solution provided by the invention

[0010] In order to remedy these drawbacks, the present invention relates, in its most general sense, to an automatic waste treatment process having the combination of technical characteristics stated in claim 1.

[0011] The excavated material includes aggregates, construction and demolition waste, excavated soil, and sediment, transported on a conveyor equipped with a hyperspectral imager. Its processing takes place on a conveyor equipped with a hyperspectral imager that includes real-time comparison capabilities for comparing the imaged excavated material flow area with a database of hyperspectral signatures characteristic of undesirable constituents. These capabilities control a means of deflecting the flow to a secondary container if undesirable constituents are detected in the imaged area (8), and / or a means of deflecting the flow to another secondary container if desirable constituents are detected in the imaged area.

[0012] Advantageously, said hyperspectral imager consists of hyperspectral sensors with a sensitivity range between 100 microns and 200 nanometers.

[0013] According to one variant, the said hyperspectral imager consists of a hyperspectral camera or a multispectral camera.

[0014] According to another variant, the said hyperspectral imager consists of an assembly of sensors forming a composite multispectral sensor.

[0015] According to another variant, the hyperspectral imager consists of an assembly of sensors with a sensitivity range between 100 microns and 200 nanometers and sensors with a sensitivity range below 200 nanometers, including X-rays.

[0016] According to a particular embodiment, hyperspectral analysis of reflection and / or photoluminescence is performed with a first set of equipment by illuminating the imaged portion with a fluorescence excitation source and with at least one spectral sensor sensitive over a spectrum ranging from thermal infrared to ultraviolet, characterized in that it comprises: a training sequence consisting of analyzing a plurality of reference samples and recording in a training database a) the spectral reflection signature acquired by the spectral analysis b) the known values ​​of the variables representing the contaminants present in each of said reference samples c) the known values ​​of the variables representing the substrates of each of said reference samples; a calibration sequence of field analysis equipment with respect to said first equipment, said field equipment comprising a light source and a spectral sensor; analysis sequences of a soil sample from a geological site consisting of acquiring the reflection signature and / or photoluminescence of said sample using said field equipment thus calibrated.and to proceed with the estimation of the pollutant characterization by processing said signature with a machine learning engine using data from the database created during the training sequence. Detailed description of a non-limiting example of an embodiment of the invention

[0017] The present invention will be described in more detail with reference to non-limiting examples of embodiments specifying the aforementioned advantages and considerations. A more particular description of the invention is briefly described below with reference to the accompanying drawings where: [ Fig.1 ] there [ Fig.1 ] represents a schematic view of a sorting installation according to the invention. General principle

[0018] There [ Fig.1 ] represents a schematic view of an installation according to the invention, intended to process excavated material from a public works site.

[0019] For the purposes of this patent, "excavated material" means mineral and / or organic materials removed from a construction site, and including, depending on the site, loose rubble from, for example, a demolition site, construction waste, excavated soil, and sediments.

[0020] It includes a rubble conveyor belt (1) carrying inert, recyclable rubble into a recovery skip (2), or, depending on the position of a hatch (4), into a waste bag or skip (3) intended to recover non-inert or non-recyclable waste.

[0021] This trapdoor (or any other form of diverting the contents of the conveyor belt) is controlled by a computer (5) running a program for processing data from a characterization cell (6).

[0022] This cell (6) is formed by a space surrounded by a tarpaulin (7) which reduces light disturbances to an imaged portion (8) of the rubble transported by the conveyor belt (1). This cell (6) includes a hyperspectral imager (10) and optionally an excitation source (11).

[0023] The aim is to detect in real time the presence of contaminants (sulfate, organic pollutant) and their nature (type of rock, hardness), and / or to classify plastic objects and possibly sort them qualitatively or quantitatively according to the type of micro-plastics and / or sort materials qualitatively or quantitatively passing through the imaged portion (8) in order to control the position of the trap (4).

[0024] The imager (10) consists of a sensor or set of hyperspectral sensors with a sensitivity range between 100 microns and 200 nanometers. It may be a hyperspectral camera, a multispectral camera, or an assembly of sensors forming a composite multispectral sensor. Additional probes or sensors

[0025] Optionally, a laser or X-ray source and corresponding sensors can be coupled with hyperspectral imaging to enhance or complement the characterization of certain materials or pollutants.

[0026] Optionally, a robotic probe (104) coupled with an optical fiber can be introduced into the cargo to obtain multispectral information at its core. The probe (104) can also be a physicochemical sensor; for example, a thermal probe, a pH meter, a sound sensor to characterize sound signatures, or a pressure sensor.

[0027] According to another variant, it further comprises at least one three-dimensional image sensor and means for processing the signals provided by said sensor for estimating the volume of said conveyor. Variant of processing from a training set

[0028] The characterization of materials present in the content passing through the imaged area (8) can be performed by algorithmic processing or by a neural network after a supervised learning phase. The predictive model is trained using a database and reference samples. A first processing step involves converting raw data from the sensor into reflection. This normalization step refers to the method that uses raw data measured on a reference material with >99% reflection (Spectralon(R)) and electronic noise data measured without an illuminant (source) to normalize the sample data between these two spectra (i.e., 0 and 100% reflection).

[0029] To eliminate these measurements prior to data acquisition, a model is generated from the prior recording of this raw reference data. Raw data measured on eight reference materials (from 2% to 99% reflectance) and electronic noise are used; a model can be trained for each combination of parameters from a device. A second level of processing predicts the variables of interest (soil composition, presence of pollutants, and pollutant quantity) based on the reflection of a sample. Several training databases are used: published training databases (pure spectral compound libraries, e.g., the USGS Spectral Library), data produced on artificial samples produced in the laboratory, or data produced on samples analyzed in the laboratory.In the case where a batch of analyzed samples comes from a particular site, a model can be trained on that batch only or that batch can be used to improve a pre-trained model based on an existing one by transfer learning method.

[0030] Data processing is applied to imaging data; data from analyzed subsamples allows for the interpretation or refinement of an initial interpretation of core samples. The model generated with spectrometer data enables real-time analyses, including analyses referenced to data from COFRAC-certified laboratories. The combination of on-site imaging and spectrometry addresses both the diagnostic phase of a site and the subsequent construction phase. Selective sorting of excavated soil based on its waste classification is also possible. Variant of the learning stage

[0031] Learning can be shared from laboratory analyses, with equipment equipped with a high-performance hyperspectral camera, to record the spectral signatures of a large number of reference samples, and provide a database accessible to a plurality of field equipment equipped with less powerful and less expensive sensors.

[0032] To account for technical and optical differences, each field instrument is calibrated using reference samples whose spectral signature has been previously recorded in the database. A correction function is then calculated to allow the database content to be used with equipment different from that used for the initial analysis.

[0033] The samples are distinguished on the one hand by the nature of the substrate, and on the other hand by the nature of the pollutants present.

[0034] The substrates are characterized by meta-descriptors based on variables such as: The chemical nature of the mineral and organic constituents, the water content, the oxide content, the pH, the particle size, the membership of one or more mineral classes according to the Strunz classification, the redox potential.

[0035] The reference substrate can be characterized by physicochemical analyses. It can also be prepared from predetermined components to prepare substrates by assembly.

[0036] Reference pollutants are characterized by their chemical composition.

[0037] Next, for each reference sample, the spectral signature is recorded by exposing it to illumination from a light source, such as a xenon lamp. The reflected light and the light emitted by photoluminescence are captured in a wavelength range from thermal infrared to ultraviolet UVC. The data are recorded for each sample along with a reference sample identifier and the physicochemical characteristics.

[0038] According to a preferred alternative, spectral acquisition of the imaged area (8) is carried out first, then a (sub-)sample (or several) is extracted for physicochemical analysis. Deconstruction materials

[0039] In one variant, the loads consist of deconstruction materials from a demolition site. The sensors make it possible to verify the nature of the waste and debris and its classification in relation to the standards for landfilling, reuse or recycling (sulfate content, bituminous coating content, etc.).

[0040] This variant also concerns the verification of loads presented at the entrance to a waste recovery site, to ensure the load conforms to the type of excavated material authorized. Rubble, secondary raw materials, sediments, and excavated soil can be inspected and sorted after transport in truck or barge containers.

Claims

1. Method for automatic processing of excavated material on a conveyor equipped with a hyperspectral imager (10) and a means (4) for deflecting the flow towards a secondary container (3) in the event that undesirable constituents are detected in said imaged zone (8), said hyperspectral imager including means for real-time comparison of the imaged zone (8) of the excavated material flow with a database of hyperspectral signatures (15) characteristic of the undesirable constituents, said real-time comparison means controlling said flow deflection means (4) towards a secondary container (3) in the event that undesirable constituents are detected in said imaged zone (8) and / or controlling said flow deflection means (4) towards another secondary container (3) in the event that desirable constituents are detected in said imaged zone (8).

2. Method for automatic processing of excavated material on a conveyor according to claim 1, characterized in that said hyperspectral imager (10) consists of hyperspectral sensors having a sensitivity range of between 100 microns and 200 nanometers.

3. Method for automatic processing of excavated material on a conveyor according to claim 1, characterized in that said hyperspectral imager (10) consists of a hyperspectral camera or a multispectral camera.

4. Method for automatic processing of excavated material on a conveyor according to claim 1, characterized in that said hyperspectral imager (10) consists of an assembly of sensors forming a composite multispectral sensor.

5. Method for automatic processing of excavated material on a conveyor according to claim 1, characterized in that said hyperspectral imager (10) consists of an assembly of sensors having a sensitivity range of between 100 microns and 200 nanometers and of sensors having a sensitivity range of less than 200 nanometers, in particular X-ray sensors.

6. Method for automatic processing of excavated material on a conveyor according to claim 1, characterized in that the hyperspectral analysis of the reflection and / or photoluminescence is carried out by means of a first piece of equipment by illuminating the imaged portion (8) by means of a fluorescence excitation source (11) and by means of at least one spectral sensor (10) sensitive over a spectrum ranging from thermal infrared to ultraviolet, characterized in that it includes: - a sequence for learning, consisting in analyzing a plurality of reference samples, and for recording the following in a learning database ∘ a) the spectral signature of reflection acquired by the spectral analysis ∘ b) known values of variables that are representative of contaminants present in each of said reference samples ∘ c) known values of variables that are representative of substrates of each of said reference samples - a sequence for calibrating a piece of field analysis equipment with respect to said first piece of equipment, said piece of field equipment including a light source and a spectral sensor, - sequences for analyzing a soil sample from a geological site, consist in acquiring the reflection and / or photoluminescence signature of said sample using said piece of field equipment thus calibrated, - and in estimating the characterization of pollutants by processing said signature by means of a learning engine that uses the data from the database created during the learning sequence.

Images